Knowledge space theory(KST)-based adaptive learning system and operating method of the same

Abstract

According to one embodiment of the present disclosure, a method of operation of a Knowledge Space Theory (KST)-based adaptive learning system is provided. The method of operation includes receiving a first user input indicating at least one sub-unit from a user device; loading knowledge items corresponding to at least one sub-unit and vectors corresponding to the knowledge items from first and second databases, respectively, wherein the vectors represent hierarchies between the knowledge items; generating a knowledge space including knowledge states, which are sets containing knowledge items, based on the knowledge items and vectors; determining a current knowledge state among the knowledge states; determining higher knowledge states having the current knowledge state as a subset; and outputting a system response recommending a learning curriculum to the user device based on the higher knowledge states.

Description

Technology Field

[0001] The present disclosure relates to an adaptive learning system and a method of operating the same, and more specifically, to a Knowledge Space Theory (KST)-based adaptive learning system and a method of operating the same. Background Technology

[0002] General learning diagnostic and recommendation systems aim to diagnose a learner's learning status and recommend learning paths based on the diagnostic results. However, conventional adaptive learning systems are generally structured to diagnose a learner's level across broad academic fields or the entire curriculum. This approach has a limitation in that it is difficult to specifically identify which sub-units require additional study.

[0003] Furthermore, existing learning diagnostic systems require learners to answer an excessive number of diagnostic questions to assess a broad range of learning, which can impose an excessive time and mental burden on them. In particular, most commercialized learning systems, both domestically and internationally, employ either a total enumeration diagnostic method (i.e., a method requiring responses to questions to evaluate the learning status across the entire subject) or are merely fixed-type tests that lack adaptive capabilities. Such methods not only involve a large number of questions but also suffer from the drawbacks of making efficient diagnosis difficult and limiting the ability to accurately infer the learning status.

[0004] Therefore, there is a growing need for adaptive learning systems that enable precise diagnosis at the sub-unit level, effectively infer a learner's status with a small number of items, and repeatedly perform precise learning recommendations and re-diagnosis based on these diagnostic results. Prior art literature

[65535] Registered Patent Publication No. 10-2293234 The problem to be solved

[0005] According to one embodiment of the present disclosure, a Knowledge Space Theory (KST)-based adaptive learning system and a method of operating the same are provided. means of solving the problem

[0006] According to one embodiment of the present disclosure, a method of operation of a Knowledge Space Theory (KST)-based adaptive learning system is provided. The method of operation comprises receiving a first user input indicating at least one sub-unit from a user device; loading knowledge items corresponding to the at least one sub-unit and vectors corresponding to the knowledge items from first and second databases, respectively, wherein the vectors represent hierarchies between the knowledge items; generating a knowledge space including knowledge states, which are sets containing the knowledge items, based on the knowledge items and the vectors; determining a current knowledge state among the knowledge states; determining higher knowledge states having the current knowledge state as a subset; and outputting a system response recommending a learning curriculum to the user device based on the higher knowledge states.

[0007] According to one embodiment of the present disclosure, a Knowledge Space Theory (KST)-based adaptive learning system communicating with a user device is provided. The KST-based adaptive learning system comprises a server including a first database and a second database, a knowledge state inference device, and a curriculum recommendation device, wherein the knowledge state inference device receives user input indicating at least one sub-unit from the user device, loads knowledge items corresponding to the at least one sub-unit and vectors corresponding to the knowledge items, respectively, generates a knowledge space including knowledge states which are sets including the knowledge items based on the knowledge items and the vectors, and determines a current knowledge state among the knowledge states, and the curriculum recommendation device determines higher knowledge states having the current knowledge state as a subset, and is configured to recommend a learning curriculum to the user device based on the higher knowledge states, and the vectors represent hierarchies between the knowledge items. Effects of the invention

[0008] According to one embodiment of the present disclosure, a KST-based adaptive learning system and a method of operating the same are provided.

[0009] In addition, a KST-based adaptive learning system and a method of operation thereof are provided, which, based on knowledge space theory, can precisely infer a learner's learning status at the level of sub-units or knowledge items using only a small number of questions, and recommend an optimal learning curriculum to the learner based on this. Brief explanation of the drawing

[0010] FIG. 1 is a block diagram illustrating a learning environment according to one embodiment of the present disclosure. FIG. 2 is a block diagram illustrating a KST-based adaptive learning system according to one embodiment of the present disclosure. FIG. 3 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to one embodiment of the present disclosure. FIG. 4 is a drawing for explaining subsections according to some embodiments of the present disclosure. FIG. 5 is a drawing for illustrating knowledge items and vectors according to some embodiments of the present disclosure. FIG. 6 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. FIG. 7 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. FIG. 8a is a drawing illustrating the creation of a knowledge space according to some embodiments of the present disclosure. FIG. 8b is a drawing illustrating a knowledge space according to some embodiments of the present disclosure. FIG. 8c is a drawing illustrating the creation of a knowledge space according to some embodiments of the present disclosure. FIG. 9a is a drawing illustrating the determination of the current state of knowledge according to some embodiments of the present disclosure. FIG. 9b is a drawing illustrating the determination of the current state of knowledge according to some embodiments of the present disclosure. FIG. 9c is a drawing illustrating the determination of the current state of knowledge according to some embodiments of the present disclosure. FIG. 10 is a drawing illustrating the determination of a higher knowledge state according to some embodiments of the present disclosure. FIG. 11 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. FIG. 12 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. Specific details for implementing the invention

[0011] In the following, embodiments of the present disclosure will be described clearly and in detail so that a person skilled in the art can easily practice the present disclosure.

[0012] Hereinafter, embodiments of the present disclosure are described in more detail with reference to the attached drawings. In describing the present disclosure, similar reference numerals are used for similar components in the drawings to facilitate overall understanding, and redundant descriptions of similar components will be omitted.

[0013] The modules in the drawings or detailed description below may be connected to components other than those shown in the drawings or described in the detailed description. Connections between modules or components may be direct or indirect. Connections between modules or components may be communication connections or physical connections.

[0014] Components described by reference to terms such as unit, module, and layer used in the detailed description may be implemented in the form of software, hardware, or a combination thereof. For example, software may be machine code, firmware, embedded code, and application software. For example, hardware may include electrical circuits, electronic circuits, processors, computers, integrated circuit cores, pressure sensors, inertial sensors, Micro Electro Mechanical Systems (MEMS), passive components, or a combination thereof.

[0015] FIG. 1 is a block diagram illustrating a learning environment according to one embodiment of the present disclosure. Referring to FIG. 1, the learning environment (LE) may include a Knowledge Space Theory (KST)-based adaptive learning system (100) and a user device (UD). The KST-based adaptive learning system (100) may be a computing system configured to process various information or to store processed information as data. In some embodiments, the KST-based adaptive learning system (100) may be implemented as a personal computer (PC), notebook, laptop, server, workstation, tablet PC, smartphone, digital camera, black box, etc.

[0016] The KST-based adaptive learning system (100) can communicate with a user device (UD). For example, the KST-based adaptive learning system (100) is connected to the user device (UD) via a wired or wireless network and can exchange input and response with the user device (UD). The KST-based adaptive learning system (100) can process the input received from the user device (UD) and output the processing result as a response to the user device (UD).

[0017] A KST-based adaptive learning system (100) can determine (or infer) a user's learning state and provide a learning curriculum to the user based on the determined learning state. For example, the KST-based adaptive learning system (100) can receive user input through a user device (UD), perform various actions to determine the user's learning state, and provide a system response to the user device (UD) recommending a learning curriculum based on the determined learning state.

[0018] In some embodiments, a single subject (e.g., first-year middle school science, second-year middle school mathematics, etc.) may include multiple major units. Additionally, a single major unit (e.g., “Electricity and Magnetism,” “Production and Transport of Electric Energy,” etc.) may include multiple minor units (e.g., “Generation of Electricity,” etc.). That is, minor units may be detailed learning units that constitute the learning curriculum.

[0019] These sub-units may consist of at least one knowledge item. For example, knowledge items may be learning objects, which are individual concepts and principles that must be acquired to study a sub-unit, which is a detailed learning unit constituting the learning curriculum. Each knowledge item may be defined as a core concept or functional unit necessary to achieve a specific learning objective within the content of the sub-unit. For instance, when the sub-unit is “Generation of Electricity,” the knowledge items of this sub-unit may be “Basic Concepts of Frictional Electricity,” “Changes in Electron Distribution Before and After Friction Between Two Objects,” etc.

[0020] The KST-based adaptive learning system (100) can determine the user's learning status in units of knowledge items and recommend a learning curriculum in units of knowledge items. The KST-based adaptive learning system (100) can specifically determine which knowledge items need to be learned to complete the learning of a specific sub-unit and provide a system that effectively recommends a learning curriculum to the user. A more detailed description of the KST-based adaptive learning system (100) will be provided later with reference to FIG. 2.

[0021] A user device (UD) may include an input device (ID) and an output device (OD). For example, the input device (ID) may include a keyboard, a touch screen, a remote control pad, a pointing device (such as a mouse, touch pad, stylus, trackball, or joystick), a scanner, a camera, or a microphone. Additionally, the output device (OD) may include a monitor display, a haptic feedback device, a television, a printer, a stereo, or a speaker.

[0023] FIG. 2 is a block diagram illustrating a KST-based adaptive learning system according to one embodiment of the present disclosure. Referring to FIG. 2, the KST-based adaptive learning system (100) may include a knowledge state inference device (110), a curriculum recommendation device (120), and a server (130).

[0024] The knowledge state inference device (110) can receive user input (UI) from a user device (UD). For example, the knowledge state inference device (110) can receive user input (UI) from the user device (UD) pointing to at least one sub-unit. The user input (UI) may point to sub-units that the user wants to learn or wants to check the learning state of (e.g., “generation of current,” “magnetic action of current,” etc.).

[0025] In some embodiments, the number of sub-units indicated by the user input (UI) may be one or more. When the number of sub-units indicated by the user input (UI) is two or more, these sub-units may be included in one major unit (e.g., “Electricity and Magnetism,” etc.) or one subject (e.g., Middle School 1st Year Science, Middle School 2nd Year Mathematics, etc.), as well as in different major units or subjects. In other words, the types of sub-units indicated by the user input (UI) (i.e., which major unit or subject they are included in) may not be limited.

[0026] The knowledge state inference device (110) can load knowledge items (KI) and vectors (VT) from the server (130). For example, the knowledge state inference device (110) can load knowledge items (KI) corresponding to sub-sections pointed to by user input (UI) and vectors (VT) corresponding to the knowledge items (KI), respectively, from the server (130) or the first and second databases (131, 132).

[0027] Knowledge Items (KIs) may be learning objects that constitute a sub-unit. For example, Knowledge Items (KIs) may be learning objects that are individual concepts and principles that must be acquired to learn a sub-unit, which is a detailed learning unit that constitutes the learning curriculum. Each Knowledge Item (KI) may be defined as a core concept or functional unit necessary to achieve a specific learning objective within the learning content of the sub-unit. For instance, when the sub-unit is “Generation of Electricity,” the Knowledge Items (KIs) of this sub-unit may be “Charging by Frictional Electricity,” “Action of Electric Force,” “Electrostatic Induction,” etc.

[0028] For example, when the knowledge state inference device (110) receives a user input (UI) pointing to the “generation of electricity” sub-unit, it may load knowledge items (KI) corresponding to the “generation of electricity” sub-unit, including “charging by triboelectricity,” “action of electric force,” “electrostatic induction,” etc., from the server (130) or the first database (131). A more detailed description of the sub-unit and knowledge items (KI) will be given later with reference to FIG. 4.

[0029] A hierarchical structure based on precedence or inclusion relationships may exist among knowledge items (KIs). For example, to learn the first knowledge item, a user may need to learn the second knowledge item first. In other words, learning the first knowledge item may imply learning the second knowledge item.

[0030] Vectors (VT) can represent a hierarchy among knowledge items (KI). For example, vectors (VT) may have a directionality that expresses the learning order, prior learning relationship, or dependency among knowledge items. For instance, one vector (VT) may have a directionality from a second knowledge item to a first knowledge item, thereby indicating that the second knowledge item is a prerequisite for the first knowledge item. A more detailed description of the vectors (VT) will be provided later with reference to FIG. 5.

[0031] The knowledge state inference device (110) can generate a knowledge space (KS). For example, the knowledge state inference device (110) can generate a knowledge space (KS) containing knowledge states based on loaded knowledge items (KI) and vectors (VT).

[0032] Knowledge states may be sets of knowledge items (KI). For example, a knowledge state inference device (110) may generate knowledge states having loaded knowledge items (KI) as elements. For example, when first to fifth knowledge items are loaded, the knowledge state inference device (110) may generate a first knowledge state having the first knowledge item as an element, a second knowledge state having the first and third knowledge items as elements, and a third knowledge state having the first to fifth knowledge states as elements.

[0033] In some embodiments, the knowledge state inference device (110) can generate all possible sets having loaded knowledge items (KI) as elements as knowledge states. For example, when three knowledge items (KI) are loaded, the knowledge state inference device (110) can generate eight knowledge states, including a knowledge state that is an empty set.

[0034] A knowledge space (KS) may include one or more knowledge states generated based on loaded knowledge items (KI). In some embodiments, the knowledge space (KS) may include the remainder of the generated knowledge states after erroneous knowledge states have been eliminated. An erroneous knowledge state may be a knowledge state in which some knowledge items are missing when a vector (VT) (or hierarchy) is considered.

[0035] For example, if the second knowledge item is a prerequisite for the first knowledge item (i.e., if the first vector points to the first knowledge item as a higher knowledge item and the second knowledge item as a lower knowledge item), a knowledge state containing only the first knowledge item may be an erroneous knowledge state. Since this implies that if a learner has acquired the first knowledge item, they have also acquired the second knowledge item as a prerequisite, a knowledge state containing only the first knowledge item and not the second knowledge item is judged to be an erroneous knowledge state and can be eliminated. Accordingly, the knowledge state inference device (110) can eliminate the erroneous knowledge state based on the vector (VT) and then generate a knowledge space (KS) as a result. A more detailed explanation of generating the knowledge space (KS) will be described later with reference to FIGS. 8a through 8c.

[0036] The knowledge state inference device (110) can determine the current knowledge state (CKST) among a plurality of knowledge states included in the knowledge space (KS). Here, the current knowledge state (CKST) is a knowledge state that includes knowledge items currently acquired (or learned) by the user among the knowledge items (KI) of the sub-unit, and can represent the user's learning state. In some embodiments, the KST-based adaptive learning system (100) can determine (or infer) the user's learning state based on the current knowledge state (CKST) and recommend an appropriate learning curriculum to the user.

[0037] The knowledge state inference device (110) can store the generated knowledge space (KS) and the determined current knowledge state (CKST) in a server (130) or a third database (133).

[0038] In some embodiments, the knowledge state inference device (110) transmits a question (QT) that diagnoses knowledge items (KI) to a user device (UD) to determine the current knowledge state (CKST), and after eliminating some knowledge states based on the user response (UR) to the question (QT) (if there is one remaining knowledge state), the remaining knowledge state can be determined as the current knowledge state (CKST).

[0039] In some embodiments, the knowledge state inference device (110) can determine a first knowledge item among the knowledge items (KI). The first knowledge item may be the knowledge item with the largest branching effect. The branching effect may be determined by determining the number of times each knowledge item is included in the knowledge states, and then determining that the knowledge item with the smallest difference between that number and half the number of knowledge states is the largest. The knowledge state inference device (110) can select the knowledge item with such a number as the first knowledge item. By selecting the knowledge item with the largest branching effect as the first knowledge item, the knowledge state inference device (110) can most stably eliminate the knowledge states depending on whether the knowledge item is learned.

[0040] For example, within a knowledge space (KS) containing 10 knowledge states, the knowledge state inference device (110) can consistently eliminate knowledge states because, depending on whether the knowledge item included in 5 knowledge states is correct or incorrect, it can eliminate 5 knowledge states when it determines that the user has not learned the knowledge item and eliminate the remaining 5 knowledge states when it determines that the user has learned it. Therefore, rather than a specific knowledge item being included in all knowledge states or only in a few knowledge states, a knowledge item included the closest number to half the total number of knowledge states can have the greatest branching effect.

[0041] In some embodiments, when there are two or more knowledge items with a large branching effect, the knowledge state inference device (110) may determine the knowledge item with the lowest hierarchy among them as the first knowledge item (with the largest branching effect). However, if the knowledge items do not have a hierarchical relationship, any knowledge item among them may be determined as the first knowledge item (with the largest branching effect).

[0042] In some embodiments, the knowledge state inference device (110) may transmit a question (QT) to a user device (UD). For example, the knowledge state inference device (110) may transmit a question (QT) that diagnoses a first knowledge item to the user device (UD). The question (QT) may include at least one of various types of questions regarding the first knowledge item (e.g., multiple choice, short answer, descriptive, essay, etc.). The question (QT) may be configured to evaluate the user's understanding of the first knowledge item, ability to apply concepts, problem-solving ability, etc., from various angles.

[0043] In some embodiments, the knowledge state inference device (110) may load a question (QT) from the server (130) or the fourth database (134). For example, the knowledge state inference device (110) may load a question (QT) corresponding to a first knowledge item from the server (130) or the fourth database (134) and transmit the loaded question (QT) to a user device (UD).

[0044] In some embodiments, the knowledge state inference device (110) may provide an index corresponding to a first knowledge item to a server (130) or a fourth database (134), and may randomly load at least one of a plurality of items (QT) corresponding to the index stored in the fourth database (134). The knowledge state inference device (110) may transmit the randomly loaded item (QT) to a user device (UD).

[0045] In some embodiments, since the items (QT) are selected randomly, the knowledge state inference device (110) can determine whether the first knowledge item has been learned based on various item configurations. Additionally, the knowledge state inference device (110) can determine whether the first knowledge item has been learned through different items (QT) each time, even for the same knowledge item. Therefore, the knowledge state inference device (110) can evaluate the user's actual learning state more precisely by preventing the user from responding to the items (QT) based on memory.

[0046] In some embodiments, the knowledge state inference device (110) may eliminate some of the knowledge states. For example, the knowledge state inference device (110) may eliminate some of the knowledge states based on the user response (UR) to the question (QT) received from the user device (UD).

[0047] In some embodiments, the knowledge state inference device (110) may erase knowledge states containing a first knowledge item or erase knowledge states not containing a first knowledge item based on the correctness of user responses (UR). A more detailed description of erasing some of the knowledge states will be given later with reference to FIGS. 9a through 9c.

[0048] The curriculum recommendation device (120) can determine higher knowledge states (UKST). For example, the curriculum recommendation device (120) can load a knowledge space (KS) and a current knowledge state (CKST) from a server (130) (or a third database (133)) and determine higher knowledge states (UKST) that have the current knowledge state (CKST) as a subset. For example, the curriculum recommendation device (120) can determine higher knowledge states (UKST) that have the current knowledge state (CKST) as a subset by adding other knowledge items (i.e., target knowledge items that the user has not yet learned) to the knowledge items within the current knowledge state (CKST). A more detailed explanation of determining the higher knowledge states (UKST) will be described later with reference to FIG. 10.

[0049] The curriculum recommendation device (120) can output a system response (SR) to the user device (UD). For example, the curriculum recommendation device (120) can generate a system response (SR) that recommends a learning curriculum based on higher knowledge states (UKST) and output the generated system response (SR) to the user device (UD).

[0050] In some embodiments, the curriculum recommendation device (120) may perform set difference operations. A set difference operation may, for example, mean an operation to derive knowledge items that are not included in the second knowledge state among the knowledge items included in the first knowledge state. For example, the curriculum recommendation device (120) may perform set difference operations between the upper knowledge states (UKST) and the current knowledge state (CKST), and determine a target knowledge item among the knowledge items in the upper knowledge states (UKST) based on the results of the set difference operations. This target knowledge item may refer to a knowledge item among the knowledge items in the knowledge space (KS) that the user has not yet learned.

[0051] The curriculum recommendation device (120) can generate a system response (SR) that recommends a learning curriculum for learning these target knowledge items and output the generated system response (SR) to a user device (UD). At this time, the learning curriculum may include learning materials of various types (e.g., video, text, interactive content, etc.) and multiple questions (QT) capable of diagnosing whether the knowledge items have been learned.

[0052] In some embodiments, the curriculum recommendation device (120) may generate a system response (SR) requesting the user to select at least one of the target knowledge items so that the user can select a learning curriculum to learn, and output it to the user device (UD). For example, the system response (SR) may include metadata regarding the learning content, learning difficulty, estimated time required, related learning objectives, etc. of each of the target knowledge items. Thus, the KST-based adaptive learning system (100) can suggest a user-customized learning path according to the user's selection.

[0053] In some embodiments, the KST-based adaptive learning system (100) may receive a user input (UI) for selecting at least one target knowledge item after outputting a system response (SR) to a user device (UD). The KST-based adaptive learning system (100) may update the current knowledge state (CKST) by diagnosing the selected target knowledge item (e.g., by generating a question (QT) that diagnoses the target knowledge item and determining whether the target knowledge item has been learned based on a user response (UR) to the question (QT).

[0054] In some embodiments, the KST-based adaptive learning system (100) may transmit questions (QT) diagnosing knowledge items (KI) in the knowledge space (KS) to a user device (UD) in response to a determination that the current knowledge state (CKST) is identical to the largest knowledge state (which may be referred to as the final knowledge state) in the knowledge space (KS) (in other words, in response to a determination that the user has learned all knowledge items (KI) in the knowledge space (KS). In other words, the KST-based adaptive learning system (100) may conduct a final formative evaluation of the knowledge space (KS) in response to a determination that the user has learned all knowledge items (KI) in the knowledge space (KS).

[0055] In some embodiments, the KST-based adaptive learning system (100) can determine the first knowledge item with the largest branching effect among the knowledge items in the knowledge space (KS) to perform a final formative evaluation of the knowledge space (KS). The knowledge state inference device (110) can provide the user device (UD) with a first question for diagnosing the first knowledge item.

[0056] Based on the correctness of the response to the first question received from the user device (UD), the knowledge state inference device (110) can eliminate some of the knowledge states in the knowledge space (KS) and determine the second knowledge item with the largest branching effect among the knowledge items included in the remaining knowledge states. The knowledge state inference device (110) can repeatedly perform the above operations until the number of remaining knowledge states in the knowledge space (KS) becomes 1.

[0057] When the number of remaining knowledge states becomes 1, the knowledge state inference device (110) determines the knowledge state as the current knowledge state (CKST) and may store it in the server (130) or the third database (133). The curriculum recommendation device (120) may load the current knowledge state (CKST) from the server (130) or the third database (133) and output a system response (SR) containing knowledge items included in the current knowledge state (CKST) to the user device (UD) as a final formative evaluation result. In some embodiments, based on the result of the final formative evaluation, the user may effectively learn knowledge items within the knowledge space (KS) by relearning knowledge items that were not accurately learned or were forgotten.

[0058] The server (130) may include first to fourth databases (131 to 134). Although the first to fourth databases (131 to 134) have been described as being included within a single server (130) by example, the scope of the disclosure is not limited thereto. In some embodiments, the first to fourth databases (131 to 134) may be included within different servers or different devices.

[0059] The first database (131) may store knowledge items (KI). For example, the first database (131) may store knowledge items (KI) of various subjects. In some embodiments, the first database (131) may store mapping information between knowledge items (KI) and sub-units. A sub-unit may include one or more knowledge items (KI), and a knowledge item (KI) may be defined as a learning objective or conceptual unit within that sub-unit. The first database (131) may store metadata including a unique identifier, name, corresponding sub-unit, etc. for each knowledge item (KI).

[0060] The second database (132) can store vectors (VT). For example, the second database (132) can store vectors (VT) representing hierarchies between knowledge items (KI). Each of the vectors (VT) can represent a directionality between a preceding knowledge item and a subsequent knowledge item, and can include information about other knowledge items that must be learned prior to learning a specific knowledge item.

[0061] The third database (133) may include a knowledge space (KS) and a current knowledge state (CKST). For example, the third database (133) may receive the knowledge space (KS) and the current knowledge state (CKST) from the knowledge state inference device (110) and store them. Additionally, the third database (133) may provide the stored knowledge space (KS) and the current knowledge state (CKST) to the curriculum recommendation device (120) upon the request of the curriculum recommendation device (120). In some embodiments, the current knowledge state (CKST) stored by the third database (133) may be updated by the knowledge state inference device (110).

[0062] The fourth database (134) may store questions (QT). For example, the fourth database (134) may store questions (QT) corresponding to each knowledge item (KI). The questions (QT) may be intended to diagnose the user's understanding of the knowledge item, ability to apply concepts, or problem-solving ability.

[0063] In some embodiments, the fourth database (134) may store metadata regarding the types of questions (QT) (e.g., multiple choice, short answer, essay, etc.), difficulty level, correct answer, question history, correct answer rate, etc.

[0065] FIG. 3 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to one embodiment of the present disclosure. Referring to FIG. 2 and FIG. 3, the KST-based adaptive learning system (100) receives user input (UI) indicating a sub-unit from a user device (UD) and can output a system response (SR) recommending a learning curriculum to the user device (UD).

[0066] In step S110, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may receive user input (UI) pointing to at least one sub-unit from a user device (UD). In some embodiments, when there are two or more sub-units pointed to by the user input (UI), the sub-units may be included in one subject, but are not necessarily limited to being included in one subject. The sub-units may be included in different subjects (e.g., middle school first-year science, middle school second-year mathematics, etc.).

[0067] In step S120, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may load knowledge items (KI) and vectors (VT), respectively, from the server (130) or the first and second databases (131, 132). For example, the knowledge state inference device (110) may load knowledge items (KI) corresponding to sub-units and vectors (VT) corresponding to the knowledge items (KI), respectively, from the server (130) or the first and second databases (131, 132).

[0068] In step S130, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can generate a knowledge space (KS). For example, the knowledge state inference device (110) can generate a knowledge space (KS) containing knowledge states (KST), which are sets containing knowledge items (KI), based on loaded knowledge items (KI) and vectors (VT).

[0069] Knowledge states (KST) may be sets of knowledge items (KI). For example, a knowledge state inference device (110) may generate knowledge states (KST) having loaded knowledge items (KI) as elements.

[0070] In some embodiments, the knowledge state inference device (110) can generate all possible sets having loaded knowledge items (KI) as elements as knowledge states (KST). For example, when three knowledge items (KI) are loaded, the knowledge state inference device (110) can generate eight knowledge states (KST), including a knowledge state that is an empty set. The knowledge state inference device (110) can generate a knowledge space (KS) by eliminating some of the generated knowledge states (KST) (e.g., erroneous knowledge states). A more detailed description of generating a knowledge space (KS) by eliminating some of the knowledge states (KST) (e.g., erroneous knowledge states) will be given later with reference to FIG. 6.

[0071] At step S140, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can determine the current knowledge state (CKST). For example, the knowledge state inference device (110) can determine the current knowledge state (CKST) among the knowledge states (KST) in the knowledge space (KS). In some embodiments, the knowledge state inference device (110) can determine the remaining knowledge state as the current knowledge state (CKST) when the number of remaining knowledge states is 1 after eliminating some of the knowledge states (KST) in the knowledge space (KS). The current knowledge state (CKST) may be a set of knowledge items (KI) of the sub-unit that the user has currently acquired (or learned).

[0072] A first knowledge item (KI1) can be determined among the knowledge items (KI). In some embodiments, the knowledge state inference device (110) can determine the knowledge item among the knowledge items (KI) that has the smallest difference from half the number of times it is included in the knowledge states (KST) (i.e., closest to half the number) as the first knowledge item (KI1). For example, when the number of knowledge states (KST) is 10, the knowledge state inference device (110) can determine the knowledge item (KI) that has the closest number to half the number, which is 5, as the first knowledge item (KI1).

[0073] In some embodiments, the knowledge state inference device (110) may determine a first knowledge item among the knowledge items (KI). The first knowledge item may be the knowledge item with the largest branching effect. The branching effect may be determined by determining the number of times each knowledge item is included in the knowledge states, and then determining that the knowledge item with the smallest difference between that number and half the number of knowledge states (KST) is the largest. The knowledge item with such a number may be selected as the first knowledge item. By selecting the knowledge item with the largest branching effect as the first knowledge item (KI1), the knowledge state inference device (110) can eliminate the knowledge states most consistently depending on whether the knowledge item is learned.

[0074] In some embodiments, the knowledge state inference device (110) may store the knowledge space (KS) and the current knowledge state (CKST) in a server (130) or a third database (133).

[0075] At step S150, the KST-based adaptive learning system (100) or the curriculum recommendation device (120) can determine higher knowledge states (UKST). For example, the curriculum recommendation device (120) can load the knowledge space (KS) and the current knowledge state (CKST) from the server (130) (or the third database (133)) and determine higher knowledge states (UKST) that have the current knowledge state (CKST) as a subset. The curriculum recommendation device (120) can determine higher knowledge states (UKST) that have the current knowledge state (CKST) as a subset by adding other knowledge items (i.e., target knowledge items that the user has not yet learned) to the knowledge items within the current knowledge state (CKST).

[0076] At step S160, the KST-based adaptive learning system (100) or the curriculum recommendation device (120) may output a system response (SR) to a user device (UD). For example, the curriculum recommendation device (120) may generate a system response (SR) that recommends a learning curriculum for learning knowledge items of higher knowledge states (UKST) or target knowledge items, and output the generated system response (SR) to the user device (UD). At this time, the learning curriculum may include learning materials of various types (e.g., video, text, interactive content, etc.) and multiple questions (QT) capable of diagnosing whether the knowledge items have been learned.

[0077] In some embodiments, the curriculum recommendation device (120) may generate a system response (SR) requesting the user to select at least one of the target knowledge items so that the user can select a learning curriculum to learn, and output it to the user device (UD). For example, the system response (SR) may include metadata regarding the learning content, learning difficulty, estimated time required, related learning objectives, etc. of each of the target knowledge items. Thus, the KST-based adaptive learning system (100) can suggest a user-customized learning path according to the user's selection.

[0078] In some embodiments, the KST-based adaptive learning system (100) may receive a user input (UI) for selecting at least one target knowledge item after outputting a system response (SR) to a user device (UD). The KST-based adaptive learning system (100) may update the current knowledge state (CKST) by diagnosing the selected target knowledge item (e.g., by generating a question (QT) that diagnoses the target knowledge item and determining whether the target knowledge item has been learned based on a user response (UR) to the question (QT).

[0080] FIG. 4 is a drawing for illustrating sub-sections according to some embodiments of the present disclosure. Referring to FIG. 4, one subject (SB) includes a plurality of major sections (HU1~HU4), each of the plurality of major sections (HU1~HU4) includes a plurality of sub-sections (SU1~SU8), and each of the plurality of sub-sections (SU1~SU8) may include a plurality of knowledge items (KI).

[0081] For the sake of understanding the present disclosure, a plurality of major units (HU1–HU4) are described by way of example with four, and a plurality of minor units (SU1–SU8) are described by way of example with two for each major unit, but the scope of the present disclosure is not limited thereto. A subject (SB) may include a different number of major units, and each major unit may include a different number of minor units containing a different number of knowledge items (KI).

[0083] FIG. 5 is a diagram illustrating knowledge items and vectors according to some embodiments of the present disclosure. Referring to FIG. 5, knowledge items (KI1–KI4) may be connected to each other by vectors (VT1–VT3). The knowledge items (KI1–KI4) have a hierarchical structure, and one knowledge item may have a hierarchical relationship with respect to another knowledge item. The hierarchical relationship may be referred to as a quasi-order.

[0084] For example, the second knowledge item (KI2) may be of a higher hierarchy than the first knowledge item (KI1). Learning the second knowledge item (KI2) may necessarily mean learning the first knowledge item (KI1). For example, learning the “basic concept of electric force” may necessarily mean learning the “change in electron distribution before and after rubbing two objects together.” Accordingly, the knowledge state inference device (110) of FIG. 2 can determine whether the first knowledge item (KI1) has been learned based on whether the second knowledge item (KI2) has been learned.

[0085] Vectors (VT1–VT3) can represent a hierarchy (or hierarchical structure) between knowledge items (KI1–KI4). For example, vectors (VT1–VT3) may have a directionality that expresses the learning order, prior learning relationship, or dependency between knowledge items (KI1–KI4). The fact that the first and second knowledge items (KI1, KI2) are connected to each other, and that the second knowledge item (KI2) has a higher hierarchy than the first knowledge item (KI1), can be indicated by the first vector (VT1).

[0086] For convenience of explanation, the vectors (VT1–VT3) are depicted as directed lines, but the scope of the present disclosure is not limited thereto. The vectors (VT1–VT3) may be implemented as pairs of ordered knowledge items (KI1–KI4). Additionally, the vectors (VT1–VT3) may be implemented in various forms that can be derived by those skilled in the art.

[0087] In some embodiments, the relationships between knowledge items (KI1–KI4) connected by vectors (VT1–VT3) can be used when generating the knowledge space (KS) of FIG. 2. For example, a knowledge state that includes a second knowledge item (KI2) among the knowledge states but does not include a first knowledge item (KI1) which is a lower hierarchy (or lower knowledge item) of the second knowledge item (KI2) may be determined to have an error when considering the first vector (VT1). Accordingly, such a knowledge state (KST) may be eliminated as an erroneous knowledge state.

[0088] Knowledge items (KI1~KI4) may have hierarchical relationships with one or more other knowledge items (in other words, they may be connected to one or more other knowledge items by vectors). For example, a second knowledge item (KI2) may be connected to the first, third, and fourth knowledge items (KI1, KI3, KI4) by vectors (VT1~VT3). In this case, the second knowledge item (KI2) may be a higher-level knowledge item of the first knowledge item (KI1) and a lower-level knowledge item of the third and fourth knowledge items (KI3, KI4).

[0089] In some embodiments, knowledge items (KI1–KI4) may correspond to different sub-units, major units, or subjects. In other words, knowledge items (KI1–KI4) corresponding to different sub-units, major units, or subjects may have hierarchical relationships with one another and may be connected by vectors.

[0091] FIG. 6 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. Referring to FIG. 2 and FIG. 6, the KST-based adaptive learning system (100) can generate a knowledge space (KS).

[0092] In step S231, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can generate knowledge states (KST). For example, the knowledge state inference device (110) can generate knowledge states (KST) that include at least some of the knowledge items (KI) loaded from the server (130) (or the first and second databases (131, 132)).

[0093] In some embodiments, the knowledge state inference device (110) can generate all possible sets having loaded knowledge items (KI) as elements as knowledge states (KST). For example, the knowledge state inference device (110) can generate 16 knowledge states (KST) by loading first to fourth knowledge items from the first database (131).

[0094] In step S232, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can determine some knowledge states (KST) that include upper knowledge items. For example, the knowledge state inference device (110) can divide loaded knowledge items (KI) into upper knowledge items and lower knowledge items by vectors (VT). The knowledge state inference device (110) can determine some knowledge states among the generated knowledge states (KST) that include knowledge items divided into upper knowledge items.

[0095] In some embodiments, a knowledge item (KI) may be both a higher knowledge item and a lower knowledge item. For example, a first knowledge item may be a higher knowledge item of a second knowledge item and a lower knowledge item of a third knowledge item. When a knowledge item is both a higher knowledge item and a lower knowledge item, the knowledge state inference device (110) may determine some knowledge states as such knowledge item as a higher knowledge item.

[0096] In step S233, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can eliminate erroneous knowledge states. For example, the knowledge state inference device can determine as erroneous knowledge states those knowledge states that do not include lower knowledge items corresponding to the higher knowledge items among some knowledge states that include higher knowledge items, and can eliminate these erroneous knowledge states.

[0097] For example, when a first knowledge item and a second knowledge item are connected by a first vector and are a sub-knowledge item and a super-knowledge item, respectively, the knowledge state inference device (110) determines the knowledge states that do not include the first knowledge item among the knowledge states that include the second knowledge item as error knowledge states, and can eliminate these error knowledge states.

[0098] In step S234, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may eliminate erroneous knowledge states. For example, the knowledge state inference device (110) may generate a knowledge space (KS) containing the remainder of the knowledge states (KST) after eliminating the erroneous knowledge states. In some embodiments, the knowledge state inference device (110) may store the generated knowledge space (KS) in a server (130) or a third database (133).

[0100] FIG. 7 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. Referring to FIG. 2 and FIG. 7, the KST-based adaptive learning system (100) can determine the current knowledge state (CKST).

[0101] In step S341, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can determine the number of times knowledge items (KI) are included in knowledge states (KST), respectively. For example, the knowledge state inference device (110) can determine the number of times knowledge items (KI) are included as elements in knowledge states (KST) within the knowledge space (KS) by counting the number of times knowledge items (KI) are included as elements in knowledge states (KST), respectively.

[0102] In step S342, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may determine the knowledge item (KI) having the smallest number of times that the difference from half the number of knowledge states (KST) is the first knowledge item (KI1). For example, the knowledge state inference device (110) may determine the knowledge item (KI) having the smallest number of times that the knowledge item (KI) is included in the knowledge states (KST) among the number of times that the knowledge item (KI) is included in the knowledge states (KST) is the first knowledge item (KI1).

[0103] The knowledge item that has the smallest difference from half the number of knowledge states (KST) among the times the knowledge item (KI) is included in the knowledge states (KST) may be the knowledge item with the largest branching effect among the knowledge items (KI1). By selecting the knowledge item with the largest branching effect as the first knowledge item, the knowledge state inference device (110) can eliminate the most knowledge states depending on whether this knowledge item is learned.

[0104] In some embodiments, when there are two or more knowledge items that have the smallest difference from half the value of the number of knowledge states (KST) (i.e., the largest branching effect), the knowledge state inference device (110) may determine the knowledge item with the lowest hierarchy among them as the first knowledge item (KI1). However, if these knowledge items do not have a hierarchical relationship, the knowledge state inference device (110) may determine any knowledge item among them as the first knowledge item (KI1).

[0105] In step S343, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may transmit a first question (QT1) for diagnosing a first knowledge item (KI1) to a user device (UD). The first question (QT1) may include at least one of various types of questions (e.g., multiple choice, short answer, descriptive, essay, etc.) regarding the first knowledge item (KI1). The first question (QT1) may be configured to evaluate the user's understanding of the first knowledge item (KI1), concept application ability, problem-solving ability, etc., from various angles.

[0106] In some embodiments, the knowledge state inference device (110) may load a question (QT) from the fourth database (134). For example, the knowledge state inference device (110) may load a first question (QT1) corresponding to a first knowledge item (KI1) from the fourth database (134) and transmit the loaded first question (QT1) to a user device (UD).

[0107] In some embodiments, the knowledge state inference device (110) may provide the fourth database (134) with an index corresponding to the first knowledge item (KI1) and randomly load at least one of a plurality of items (QT) corresponding to the index stored in the fourth database (134) as the first item (QT1). The knowledge state inference device (110) may transmit the randomly loaded first item (QT1) to the user device (UD).

[0108] In step S344, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can determine whether the first user response (UR1) is correct. For example, the knowledge state inference device (110) receives the first user response (UR1) for the first question (QT1) from the user device (UD) and can determine whether the first user response (UR1) is correct for the first question (QT1).

[0109] In step S345-1, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may determine some knowledge states that include the first knowledge item (KI1). For example, the knowledge state inference device (110) may determine some knowledge states that include the first knowledge item (KI1) among the knowledge states (KST) in response to the determination that the first user response (UR1) is incorrect.

[0110] In step S345-2, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may determine some knowledge states that do not include the first knowledge item (KI1). For example, the knowledge state inference device (110) may determine some knowledge states among the knowledge states (KST) that do not include the first knowledge item (KI1) in response to the first user response (UR1) being determined to be correct.

[0111] In step S346, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may eliminate some determined knowledge states. For example, the knowledge state inference device (110) may eliminate some determined knowledge states depending on whether the first user response (UR1) is correct. When the first user response (UR1) is correct, the knowledge state inference device (110) may eliminate some knowledge states that do not include the first knowledge item (KI1). Additionally, when the first user response (UR1) is incorrect, the knowledge state inference device (110) may eliminate some knowledge states that include the first knowledge item (KI1).

[0112] In step S347, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can determine whether the number of remaining knowledge states is 1. For example, the knowledge state inference device (110) can determine whether the number of remaining knowledge states is 1 as a result of eliminating some knowledge states among the knowledge states (KST) based on whether the first user response (UR1) is correct.

[0113] In step S348, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may determine the residual knowledge state as the current knowledge state (CKST). For example, the knowledge state inference device (110) may determine the residual knowledge state as the current knowledge state (CKST) in response to determining that the number of residual knowledge states is 1. In other words, the knowledge state inference device (110) may determine the last remaining residual knowledge state among the knowledge states (KST) as the current knowledge state (CKST) that indicates the user's learning state. The knowledge items (KI) included in the current knowledge state (CKST) may indicate knowledge items that the user has learned or knows.

[0114] In some embodiments, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may repeat steps S341 through S347 in response to a determination that the number of remaining knowledge states is not 1 (i.e., the number of remaining knowledge states is 2 or more). For example, in response to a determination that the number of remaining knowledge states is not 1, the knowledge state inference device (110) may determine the number of times knowledge items (KI) are included in the remaining knowledge states (step S341), and determine the knowledge item having the smallest difference from half the number of knowledge states among the counts as a new first knowledge item (KI1) (or second knowledge item) (step S342). The knowledge state inference device (110) transmits a new first question (QT1) (or second question) to the user device (UD) to diagnose a new first knowledge item (KI1) (or second knowledge item) (step S343), determines some knowledge states based on whether the new first user response (UR1) (or second user response) is correct (step S344) (step S345), and can eliminate them (step S346). The knowledge state inference device (110) can determine whether the number of remaining knowledge states is 1 (step S347).

[0115] In some embodiments, the knowledge state inference device (110) may store the current knowledge state (CKST) in a server (130) or a third database (133).

[0117] FIGS. 8a to 8c are drawings illustrating the creation of a knowledge space according to some embodiments of the present disclosure.

[0118] FIG. 8a is a drawing illustrating the creation of a knowledge space according to some embodiments of the present disclosure. Referring to FIG. 2, FIG. 6, and FIG. 8a, a knowledge state inference device (110) may create knowledge states (KST1–KST16) based on knowledge items (KI1–KI4) and vectors (VT) loaded from a server (130) before creating a knowledge space (KS). For convenience of explanation, the first to fourth knowledge items (KI1–KI4) are illustrated exemplarily, but the scope of the present disclosure is not limited thereto.

[0119] The knowledge state inference device (110) can generate first to 16 knowledge states (KST1 to KST16), which are all possible combinations of first to fourth knowledge items (KI1 to KI4). For example, the first knowledge state (KST1) may be an empty set, and the second to fifth knowledge states (KST2 to KST5) may each include one of the first to fourth knowledge items (KI1 to KI4). The sixth to eleventh knowledge states (KST6 to KST11) may each include two of the first to fourth knowledge items (KI1 to KI4). The twelfth to fifteenth knowledge states (KST12 to KST15) may each include three of the first to fourth knowledge items (KI1 to KI4). The sixteenth knowledge state (KST16) may include all of the first to fourth knowledge items (KI1 to KI4).

[0121] FIG. 8b is a diagram illustrating the creation of a knowledge space according to some embodiments of the present disclosure. Referring to FIG. 2, FIG. 6, and FIG. 8b, the knowledge state inference device (110) may eliminate some of the created knowledge states (KST1–KST16) before creating the knowledge space (KS). For example, the knowledge state inference device (110) may determine erroneous knowledge states among the knowledge states (KST1–KST16) and eliminate the erroneous knowledge states. An erroneous knowledge state may be a knowledge state in which some knowledge items are missing when a vector (VT) (or hierarchy) is considered. For convenience of explanation, the erroneous knowledge states among the knowledge states (KST1–KST16) will be determined below by considering the vectors (VT1–VT3) of FIG. 5.

[0122] The knowledge state inference device (110) can determine upper knowledge items and lower knowledge items among the first to fourth knowledge items (KI1~KI4). For example, referring to FIG. 5, the first knowledge item (KI1) may be a lower knowledge item of the second knowledge item (KI2). The second knowledge item (KI2) may be an upper knowledge item of the first knowledge item (KI1) and a lower knowledge item of the third and fourth knowledge items (KI3, KI4). The third knowledge item (KI3) may be an upper knowledge item of the second knowledge item (KI2). The fourth knowledge item (KI4) may be an upper knowledge item of the second knowledge item (KI2). Accordingly, the knowledge state inference device (110) can determine the second to fourth knowledge items (KI2~KI4) as upper knowledge items.

[0123] The knowledge state inference device (110) can determine some knowledge states including upper knowledge items among the first to 16 knowledge states (KST1~KST16). For example, the knowledge state inference device (110) can determine the third to 16 knowledge states (KST3~KST16) including at least one of the second to fourth knowledge items (KI2~KI4) among the first to 16 knowledge states (KST1~KST16).

[0124] The knowledge state inference device (110) can determine the knowledge states among the third to sixth knowledge states (KST3 to KST16) that do not include lower knowledge items corresponding to upper knowledge items as erroneous knowledge states. For example, the third to fifth knowledge states (KST3 to KST5) each include upper knowledge items, namely the second to fourth knowledge items (KI2 to KI4), but do not include their lower knowledge items (e.g., the first and second knowledge items (KI1, KI2)). Therefore, the knowledge state inference device (110) can determine the third to fifth knowledge states (KST3 to KST5) as erroneous knowledge states. The 7th to 11th knowledge states (KST7~KST11) include at least one upper knowledge item but do not include at least one of their lower knowledge items (e.g., the 1st and 2nd knowledge items (KI1, KI2)), so the knowledge state inference device (110) can determine the 7th to 11th knowledge states (KST7~KST11) as erroneous knowledge states. The 14th and 15th knowledge states (KST14, KST15) include at least one upper knowledge item but do not include at least one of their lower knowledge items (e.g., the 1st and 2nd knowledge items (KI1, KI2)), so the knowledge state inference device (110) can determine the 14th and 15th knowledge states (KST14, KST15) as erroneous knowledge states.

[0125] The knowledge state inference device (110) can erase erroneous knowledge states. For example, the knowledge state inference device (110) can erase the erroneous knowledge states among the 1st to 16th knowledge states (KST1 to KST16), namely the 3rd to 5th knowledge states (KST3 to KST5), the 7th to 11th knowledge states (KST7 to KST11), the 14th knowledge state (KST14), and the 15th knowledge state (KST15).

[0127] FIG. 8c is a drawing illustrating the creation of a knowledge space according to some embodiments of the present disclosure. Referring to FIG. 2, FIG. 6, and FIG. 8c, a knowledge state inference device (110) can create a knowledge space (KS) that includes the remainder of the knowledge states after removing erroneous knowledge states. For example, the knowledge state inference device (110) can create a knowledge space (KS) that includes the first, second, sixth, twelfth, thirteenth, and sixteenth knowledge states (KST1, KST2, KST6, KST12, KST13, KST16) that are the remainder of the knowledge states including the first to fourth knowledge items (KI1~KI4) after removing erroneous knowledge states.

[0128] In some embodiments, the knowledge state inference device (110) may store the knowledge space (KS) in a server (130) or a third database (133).

[0130] FIGS. 9a to 9c are drawings illustrating the determination of the current knowledge state.

[0131] FIG. 9a is a diagram illustrating the determination of a current knowledge state according to some embodiments of the present disclosure. Referring to FIG. 2, FIG. 7, and FIG. 9a, a knowledge state inference device (110) determines the knowledge item with the greatest branching effect among the knowledge items and can eliminate some knowledge states depending on whether the user's response to a question diagnosing the knowledge item is correct. For convenience of explanation, the knowledge space (KS) of FIG. 8c is illustrated as an example, but the scope of the present disclosure is not limited thereto.

[0132] The knowledge state inference device (110) can determine the knowledge item with the largest branching effect among the first to fourth knowledge items (KI1~KI4). For example, the knowledge state inference device (110) can determine the number of times each of the first to fourth knowledge items (KI1~KI4) is included in knowledge states within the knowledge space (KS), and determine (or select) the knowledge item whose number has the smallest difference from half the value of the number of knowledge states (i.e., closest to half the value). In some embodiments, when there are two or more of these knowledge items, the knowledge state inference device (110) can determine (or select) the knowledge item with the lowest hierarchy among these knowledge items. The knowledge state inference device (110) can minimize the overhead for determining the current knowledge state (CKST) by determining the knowledge item with the lowest hierarchy as the knowledge item with the largest branching effect.

[0133] For example, for the first, second, sixth, twelfth, thirteenth, and sixteenth knowledge states (KST1, KST2, KST6, KST12, KST13, KST16), the first knowledge item (KI1) may be included 5 times, the second knowledge item (KI2) 4 times, the third knowledge item (KI3) 2 times, and the fourth knowledge item (KI4) 2 times. Since there are 6 knowledge states, the knowledge items that have the smallest difference from the half value of 3 (i.e., the value closest to 3) may be the second to fourth knowledge items (KI2~KI4). Thus, the second knowledge item (KI2), which has the smallest hierarchy among the second to fourth knowledge items (KI2~KI4), may have the largest branching effect. Since the second knowledge item (KI2) has the largest branching effect, like the first knowledge item (KI1) of FIG. 7, the knowledge state inference device (110) can determine (or select) the second knowledge item (KI2) among the first to fourth knowledge items (KI1~KI4).

[0134] The knowledge state inference device (110) transmits a question (QT) for diagnosing a second knowledge item (KI2) to a user device (UD), and can eliminate some of the knowledge states in the knowledge space (KS) according to the user response (UR) to the question (QT) received from the user device (UD).

[0135] The knowledge state inference device (110) can eliminate knowledge states containing the second knowledge item (KI2) in response to the determination that the user response (UR) to the question (QT) diagnosing the second knowledge item (KI2) is incorrect. For example, the knowledge state inference device (110) can eliminate the 6th, 12th, 13th, and 16th knowledge states (KST6, KST12, KST13, KST16) containing the second knowledge item (KI2).

[0136] The knowledge state inference device (110) can determine whether the number of remaining knowledge states in the knowledge space (KS) is 1 after erasing knowledge states including the second knowledge item (KI2). For example, after the knowledge state inference device (110) erases the 6th, 12th, 13th, and 16th knowledge states (KST6, KST12, KST13, KST16), the first and second knowledge states (KST1, KST2) may exist in the knowledge space (KS). Accordingly, the knowledge state inference device (110) can determine that the number of remaining knowledge states is not 1 (i.e., 2 or more).

[0138] FIG. 9b is a diagram illustrating the determination of a current knowledge state according to some embodiments of the present disclosure. Referring to FIG. 2, FIG. 7, FIG. 9a, and FIG. 9b, the knowledge state inference device (110) may additionally eliminate some of the remaining knowledge states in response to determining that the number of remaining knowledge states is not 1.

[0139] The knowledge state inference device (110) can determine the first knowledge item (KI1) that has the largest branching effect among the knowledge items included in the first and second knowledge states (KST1, KST2), which are the remaining knowledge states.

[0140] The knowledge state inference device (110) transmits a question (QT) diagnosing a first knowledge item (KI1) to a user device (UD), and can erase some knowledge states depending on whether the user response (UR) to the question (QT) received from the user device (UD) is correct. For example, in response to the knowledge state inference device (110) determining that the user response (UR) corresponding to the first knowledge item (KI1) is correct, the knowledge state inference device (110) can erase a first knowledge state (KST1) that does not include the first knowledge item (KI1).

[0141] The knowledge state inference device (110) can determine whether the number of remaining knowledge states is 1 after erasing some knowledge states based on whether the user response (UR) is correct. The knowledge state inference device (110) can determine that the number of remaining knowledge states (e.g., second knowledge states (KST2)) is 1 after erasing the first knowledge state (KST1) that does not include the first knowledge item (KI1).

[0142] The knowledge state inference device (110) may determine the remaining knowledge state as the current knowledge state (CKST) in response to the determination that the number of remaining knowledge states is 1. For example, the knowledge state inference device (110) may determine the second knowledge state (KST2) as the current knowledge state (CKST). The current knowledge state (CKST) may indicate the user's learning state. For example, the current knowledge state (CKST) may indicate that the user has learned or knows the first knowledge item (KI1) among the first to fourth knowledge items (KI1~KI4).

[0143] In some embodiments, the knowledge state inference device (110) may store the current knowledge state (CKST) in a server (130) or a third database (133).

[0145] FIG. 9c is a diagram illustrating the determination of a current knowledge state according to some embodiments of the present disclosure. Referring to FIG. 2, FIG. 7, and FIG. 9c, a knowledge state inference device (110) determines the knowledge item with the greatest branching effect among the knowledge items and can eliminate some knowledge states depending on whether the user's response to a question diagnosing the knowledge item is correct. For convenience of explanation, the knowledge space (KS) of FIG. 8c is illustrated as an example, but the scope of the present disclosure is not limited thereto.

[0146] The knowledge state inference device (110) can determine the knowledge item with the largest branching effect among the first to fourth knowledge items (KI1~KI4). For example, the knowledge state inference device (110) can determine the number of times each of the first to fourth knowledge items (KI1~KI4) is included in the knowledge states within the knowledge space (KS), and determine (or select) the knowledge item having the smallest number of times that is different from half the value of the number of knowledge states. As an example, the knowledge state inference device (110) can determine (or select) the second knowledge item (KI2) among the first to fourth knowledge items (KI1~KI4).

[0147] The knowledge state inference device (110) transmits a question (QT) for diagnosing a second knowledge item (KI2) to a user device (UD), and can eliminate some of the knowledge states in the knowledge space (KS) according to the user response (UR) to the question (QT) received from the user device (UD).

[0148] The knowledge state inference device (110) can eliminate knowledge states that do not include the second knowledge item (KI2) in response to the determination that the user response (UR) to the question (QT) diagnosing the second knowledge item (KI2) is correct. For example, the knowledge state inference device (110) can eliminate first and second knowledge states (KST1, KST2) that do not include the second knowledge item (KI2).

[0149] The knowledge state inference device (110) can determine whether the number of remaining knowledge states in the knowledge space (KS) is 1 after eliminating knowledge states that do not include the second knowledge item (KI2). For example, after the knowledge state inference device (110) eliminates the first and second knowledge states (KST1, KST2), the 6th, 12th, 13th, and 16th knowledge states (KST6, KST12, KST13, KST16) may exist in the knowledge space (KS). Accordingly, the knowledge state inference device (110) can determine that the number of remaining knowledge states is not 1 (i.e., 2 or more).

[0150] In some embodiments, the knowledge state inference device (110) may further erase some of the knowledge states among the 6th, 12th, 13th, and 16th knowledge states (KST6, KST12, KST13, KST16) until the number of remaining knowledge states becomes 1.

[0152] FIG. 10 is a diagram illustrating the determination of a higher knowledge state according to some embodiments of the present disclosure. Referring to FIG. 2 and FIG. 8c and FIG. 10, a curriculum recommendation device (120) can determine a higher knowledge state (UKST). For example, the curriculum recommendation device (120) can load a knowledge space (KS) and a current knowledge state (CKST) from a server (130) (or a third database (133)) and determine higher knowledge states (UKST) that have the current knowledge state (CKST) as a subset.

[0153] For convenience of explanation, the knowledge space (KS) of FIG. 8c is illustrated and the second knowledge state (KST2) is exemplarily described as the current knowledge state (CKST), but the scope of the present disclosure is not limited thereto.

[0154] The curriculum recommendation device (120) can determine a higher knowledge state (UKST) that has the current knowledge state (CKST) as a subset by adding other knowledge items that the user has not yet learned to the knowledge items included in the current knowledge state (CKST). The other knowledge items that the user has not yet learned may be referred to as target knowledge items.

[0155] For example, the curriculum recommendation device (120) can determine the sixth knowledge state (KST6) as the upper knowledge state (UKST) among the knowledge states of the knowledge space (KS), which has one more knowledge item than the second knowledge state (KST2), which is the current knowledge state (CKST). In other words, the curriculum recommendation device (120) can determine the sixth knowledge state (KST6), which has the second knowledge state (KST2) as a subset and additionally includes one target knowledge item, as the upper knowledge state (UKST).

[0157] FIG. 11 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. Referring to FIG. 2 and FIG. 11, the KST-based adaptive learning system (100) receives user input (UI) indicating a sub-unit from a user device (UD), generates a diagnostic knowledge space (DKS) and a learning knowledge space (EKS), and outputs a system response (SR) recommending a learning curriculum to the user device (UD).

[0158] In step S410, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may receive user input (UI) pointing to at least one sub-unit from a user device (UD). In some embodiments, when there are two or more sub-units pointed to by the user input (UI), the sub-units may be included in one subject, but are not necessarily limited to being included in one subject. The sub-units may be included in different subjects (e.g., middle school first-year science, middle school second-year mathematics, etc.).

[0159] In step S420, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may load knowledge items and vectors from the server (130). Step S420 may include steps S421 through S424.

[0160] In step S421, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may load learning knowledge items (EKI) from the server (130) or the first database (131). The learning knowledge items (EKI) may be knowledge items corresponding to at least one sub-unit pointed to by user input (UI).

[0161] In step S422, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may load learning vectors (EVT) and first diagnostic vectors (DVT1) from the server (130) or the second database (132). The learning vectors (VT) may be vectors representing hierarchies between learning knowledge items (EKI), and the first diagnostic vectors (DVT1) may be vectors in which knowledge items other than the learning knowledge items (EKI) point to the learning knowledge items (EKI) as a higher hierarchy.

[0162] In step S423, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may load diagnostic knowledge items (DKI) that the first diagnostic vectors (DVT1) point to as a lower hierarchy from the server (130) or the first database (131). The diagnostic knowledge items (DKI) may not correspond to at least one sub-unit pointed to by the user input (UI), but may correspond to other sub-units or other subjects. In other words, the diagnostic knowledge items (DKI) may be knowledge items that need to be learned prior to learning the sub-unit (or learning knowledge items (EKI)) selected by the user for learning.

[0163] In step S424, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can load second diagnostic vectors (DVT2) pointing to a higher hierarchy of diagnostic knowledge items (DKI) from the server (130) or the second database (132).

[0164] In some embodiments, the second diagnostic vectors (DVT2) may include vectors representing hierarchies between loaded diagnostic knowledge items (DKI) and vectors in which sub-knowledge items of diagnostic knowledge items (DKI) that are not yet loaded point to the diagnostic knowledge items (DKI) as a higher hierarchy.

[0165] In some embodiments, the knowledge state inference device (110) may additionally load sub-knowledge items of diagnostic knowledge items (DKI) that have not yet been loaded from the server (130) or the first database (131), and may additionally load second diagnostic vectors (DVT2) that point to these sub-knowledge items as a higher hierarchy from the server (130) or the second database (132).

[0166] Additionally, the knowledge state inference device (110) may repeatedly perform the above operations until there are no longer any sub-knowledge items of the diagnostic knowledge items (DKI) that have not yet been loaded (e.g., sub-knowledge items of sub-knowledge items) and corresponding second diagnostic vectors (DVT2). That is, steps S423 and S424 should be understood as comprising a series of operations repeated until there are finally no longer any sub-hierarchies of the diagnostic knowledge items (DKI) that have not yet been loaded and second diagnostic vectors (DVT2). In step S430, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may generate knowledge spaces. Step S430 may include steps S431 and S432.

[0167] In step S431, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may generate a learning knowledge space (EKS). For example, the knowledge state inference device (110) may generate a learning knowledge space (EKS) based on learning knowledge items (EKI), diagnostic knowledge items (DKI), learning vectors (EVT), first diagnostic vectors (DVT1), and second diagnostic vectors (DVT2). The learning knowledge space (EKS) may include learning knowledge items (EKI) corresponding to at least one sub-unit pointed to by user input (UI) and diagnostic knowledge items (DKI) that need to be learned prior to.

[0168] In some embodiments, the knowledge state inference device (110) may generate knowledge states (KST) including learning knowledge items (EKI) and diagnostic knowledge items (DKI), and generate a learning knowledge space (EKS) by eliminating erroneous knowledge states among the knowledge states (KST) based on learning vectors (EVT), first diagnostic vectors (DVT1), and second diagnostic vectors (DVT2). The knowledge state inference device (110) may store the learning knowledge space (EKS) in a server (130) or a third database (133).

[0169] In step S432, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may generate a diagnostic knowledge space (DKS). For example, the knowledge state inference device (110) may generate a diagnostic knowledge space (DKS) based on diagnostic knowledge items (DKI) and second diagnostic vectors (DVT2). The diagnostic knowledge space (DKS) may include diagnostic knowledge items (DKI) that need to be learned prior to learn learning knowledge items (EKI) corresponding to at least one sub-unit pointed to by user input (UI).

[0170] In some embodiments, the knowledge state inference device (110) can generate knowledge states (KST) including diagnostic knowledge items (DKI) and generate a diagnostic knowledge space (DKS) by eliminating erroneous knowledge states among the knowledge states (KST) based on second diagnostic vectors (DVT2).

[0171] In step S440, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can determine the current knowledge state (CKST). For example, the knowledge state inference device (110) can determine the current knowledge state (CKST) among the knowledge states (KST) within the diagnostic knowledge space (DKS).

[0172] In some embodiments, the knowledge state inference device (110) can eliminate some knowledge states (KST) within the diagnostic knowledge space (DKS) based on the user response (UR) to the question (QT) corresponding to the knowledge item with the largest branching effect among the diagnostic knowledge items (DKI). When the number of remaining knowledge states becomes one, the knowledge state inference device (110) can determine the remaining knowledge state as the current knowledge state (CKST).

[0173] In some embodiments, the knowledge state inference device (110) may store the current knowledge state (CKST) in a server (130) or a third database (133).

[0174] At step S450, the KST-based adaptive learning system (100) or the curriculum recommendation device (120) can determine the upper knowledge state (UKST). The curriculum recommendation device (120) loads the learning knowledge space (EKS) and the current knowledge state (CKST) from the server (130) or the third database (133), and can determine the upper knowledge states (UKST) that have the current knowledge state (CKST) as a subset.

[0175] The curriculum recommendation device (120) can determine higher knowledge states (UKST) that have the current knowledge state (CKST) as a subset by adding other knowledge items (i.e., target knowledge items that the user has not yet learned) to the knowledge items within the current knowledge state (CKST). In some embodiments, when the current knowledge state (CKST) is determined to be one of the knowledge states (KST) within the diagnostic knowledge space (DKS), the current knowledge state (CKST) may include only diagnostic knowledge items (DKI).

[0176] In step S460, the KST-based adaptive learning system (100) or the curriculum recommendation device (120) may output a system response (SR) to the user device (UD). For example, the curriculum recommendation device (120) may generate a system response (SR) that recommends a learning curriculum for learning knowledge items of higher knowledge states (UKST) or target knowledge items, and output the generated system response (SR) to the user device (UD).

[0177] In some embodiments, the curriculum recommendation device (120) may perform difference operations. For example, the curriculum recommendation device (120) may perform difference operations between upper knowledge states (UKST) and current knowledge state (CKST), and determine target knowledge items among the knowledge items within the upper knowledge states (UKST) based on the results of the difference operations. The target knowledge items may refer to knowledge items within the learning knowledge space (DKS) that the user has not yet learned (e.g., learning knowledge items (EKI) or diagnostic knowledge items (DKI)).

[0178] In some embodiments, the curriculum recommendation device (120) may output a system response (SR) requesting the user device (UD) to select at least one of the target knowledge items. The curriculum recommendation device (120) may determine the target knowledge items among the knowledge items in the upper knowledge states (UKST) based on the results of the difference set operations, and output a system response (SR) requesting the user device (UD) to select at least one of the target knowledge items.

[0180] FIG. 12 is a flowchart illustrating the operation method of a KST-based adaptive learning system according to some embodiments of the present disclosure. Referring to FIG. 2, FIG. 11, and FIG. 12, the KST-based adaptive learning system (100) can transmit questions (QT) to a user device (UD) for diagnosing knowledge items (e.g., learning knowledge items (EKI) and diagnostic knowledge items (DKI)) within the final knowledge state of a learning knowledge space (EKS).

[0181] In step S510, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may receive a second user input (UI2). For example, the knowledge state inference device (110) may receive a second user input (UI2) from a user device (UD) for selecting a first target knowledge item among the target knowledge items. The target knowledge items may refer to knowledge items within the learning knowledge space (EKS) that the user has not yet learned (e.g., learning knowledge items (EKI) or diagnostic knowledge items (DKI)).

[0182] In step S520, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may transmit a learning curriculum and a second question (QT2) to a user device (UD). For example, the knowledge state inference device (110) may sequentially transmit a learning curriculum for learning a first target knowledge item and a second question (QT2) for diagnosing the target knowledge item to the user device (UD). The learning curriculum may include learning materials of various types (e.g., video, text, interactive content, etc.) and multiple questions capable of diagnosing whether the corresponding target knowledge items have been learned.

[0183] At step S530, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may update the current knowledge state (CKST). For example, the knowledge state inference device (110) may receive a second user response to a second question (QT2) from a user device (UD) and update the current knowledge state (CKST) based on the second user response. For example, the knowledge state inference device (110) may update the current knowledge state (CKST) to include the first target knowledge item in response to determining that the second user response is correct. In some embodiments, the updated current knowledge state (CKST) may be stored in a server (130) or a third database (133).

[0184] In step S540, the KST-based adaptive learning system (100) or the knowledge state inference device (110) can determine whether the updated current knowledge state (CKST) is the same as the final knowledge state (FKST). Here, the final knowledge state (FKST) may be a knowledge state containing the most learning knowledge items (EKI) among the learning knowledge states. In other words, the final knowledge state (FKST) may mean a state in which all learning knowledge items (EKI) corresponding to at least one sub-unit initially selected by the user have been learned. For example, the knowledge state inference device (110) can determine whether the updated current knowledge state (CKST) contains all knowledge items within the final knowledge state (FKST) (e.g., diagnostic knowledge items (DKI) and learning knowledge items (EKI)).

[0185] In step S550, the KST-based adaptive learning system (100) or the knowledge state inference device (110) may transmit questions (QT) to the user device (UD). For example, in response to a determination that the updated current knowledge state (CKST) is identical to the final knowledge state (FKST), the knowledge state inference device (110) may transmit questions (QT) to the user device (UD) for diagnosing knowledge items (e.g., diagnostic knowledge items (EKI) and learning knowledge items) within the learning knowledge space (EKS). In other words, the KST-based adaptive learning system (100) may conduct a final formative evaluation of the learning knowledge space (EKS) in response to a determination that the user has learned all knowledge items (KI) within the learning knowledge space (EKS).

[0186] In some embodiments, the knowledge state inference device (110) can determine whether the user has sufficiently or completely learned the sub-unit originally selected by the user device (UD) by transmitting questions (QT) that diagnose knowledge items (e.g., diagnostic knowledge items (EKI) and learning knowledge items (EKI)) within the learning knowledge space (EKS). The questions (QT) may include various types such as multiple choice, short answer, descriptive, and essay types, and may be configured to evaluate the user's understanding, ability to apply concepts, problem-solving skills, etc., from various angles.

[0187] In some embodiments, the knowledge state inference device (110) may determine a first knowledge item with the largest branching effect among the knowledge items (e.g., diagnostic knowledge items (DKI) and learning knowledge items) within the learning knowledge space (EKS) to perform a final formative evaluation of the learning knowledge space (EKS). The knowledge state inference device (110) may provide a first item for diagnosing the first knowledge item to the user device (UD).

[0188] Based on the correctness of the response to the first question received from the user device (UD), the knowledge state inference device (110) can eliminate some of the knowledge states in the learning knowledge space (EKS) and determine the second knowledge item with the largest branching effect among the knowledge items included in the remaining knowledge states. The knowledge state inference device (110) can repeatedly perform the above operations until the number of remaining knowledge states in the learning knowledge space (EKS) becomes 1.

[0189] When the number of remaining knowledge states becomes 1, the knowledge state inference device (110) determines the knowledge state as the current knowledge state (CKST) and may store it in the server (130) or the third database (133). The curriculum recommendation device (120) may load the current knowledge state (CKST) from the server (130) or the third database (133) and output a system response (SR) containing knowledge items included in the current knowledge state (CKST) to the user device (UD) as a final formative assessment result. In some embodiments, the knowledge state inference device (110) may load items (QT) from the server (130) or the fourth database (134). For example, the knowledge state inference device (110) can load questions (QT) corresponding to knowledge items (e.g., learning knowledge items (EKI) and diagnostic knowledge items (DKI)) in the learning knowledge space (EKS) from the fourth database (134), and transmit the loaded questions (QT) to the user device (UD).

[0190] In some embodiments, the knowledge state inference device (110) provides the fourth database (134) with an index corresponding to the knowledge item with the greatest branching effect among the knowledge items in the learning knowledge space (EKS), and can randomly load some of the multiple questions (QT) corresponding to the index stored in the fourth database (134). The knowledge state inference device (110) can transmit the randomly loaded questions (QT) to the user device (UD). Because the questions (QT) are randomly selected, the knowledge state inference device (110) can prevent the user from responding to the questions (QT) by memorization alone.

[0191] In some embodiments, in response to a determination that the updated current knowledge state (CKST) is not identical to the final knowledge state (FKST), the knowledge state inference device (110) may repeat steps S510 through S540. That is, if there is another target knowledge item (or learning knowledge item) that is not included in the current knowledge state (CKST) within the learning knowledge space (EKS), the knowledge state inference device (110) may receive user input (UI) from a user device (UD) to select the corresponding target knowledge item.

[0193] The above descriptions are specific embodiments for carrying out the present disclosure. The present disclosure will include not only the embodiments described above, but also embodiments that can be simply modified or easily modified. Furthermore, the present disclosure will include technologies that can be easily modified and implemented using the embodiments. Accordingly, the scope of the present disclosure should not be limited to the embodiments described above, but should be defined by the claims set forth below as well as equivalents to the claims of this disclosure. Explanation of the symbols

[0194] 100: KST-based Adaptive Learning System 110: Knowledge State Inference Device 120: Curriculum Recommendation Device 130: Server

Claims

Claim 1 A method of operation of a Knowledge Space Theory (KST)-based adaptive learning system comprising a knowledge state inference device and a curriculum recommendation device, each comprising at least one processor, and a server communicating with the knowledge state inference device and the curriculum recommendation device, wherein the knowledge state inference device receives a first user input pointing to at least one learning unit from a user device; wherein the knowledge state inference device loads, respectively, a plurality of knowledge items corresponding to the at least one learning unit and a plurality of vectors representing the hierarchical relationship between the plurality of knowledge items from the server; wherein the knowledge state inference device generates sets having the plurality of knowledge items as elements as a plurality of candidate knowledge states; wherein the knowledge state inference device determines, based on the plurality of vectors, a candidate knowledge state that includes a higher-level knowledge item and does not include a lower-level knowledge item of the higher-level knowledge item as an error knowledge state, and thereby generates a knowledge space including a plurality of knowledge states by eliminating the error knowledge state; wherein the knowledge state inference device sets the plurality of knowledge states included in the knowledge space as residual knowledge states; and wherein the knowledge state inference device determines the knowledge item with the greatest branching effect on the residual knowledge states as the diagnosis target knowledge A step of determining an item, wherein the knowledge item to be diagnosed is selected such that the number of elivable knowledge states among the remaining knowledge states is maximized as the learning status of the knowledge item to be diagnosed is determined; a step of transmitting a question corresponding to the knowledge item to be diagnosed to the user device by the knowledge state inference device; a step of receiving a user response to the question from the user device by the knowledge state inference device;A method of operation comprising: a step of, by the knowledge state inference device, eliminating knowledge states among the remaining knowledge states that do not include the knowledge item to be diagnosed in response to the user response being correct, and eliminating knowledge states among the remaining knowledge states that include the knowledge item to be diagnosed in response to the user response being incorrect; a step of determining the user's current knowledge state based on the remaining knowledge states eliminated according to the user response by the knowledge state inference device; a step of determining at least one target knowledge item that can be added to the current knowledge state based on the current knowledge state by the curriculum recommendation device; and a step of outputting a system response that recommends a learning curriculum for learning the target knowledge item to the user device by the curriculum recommendation device. Claim 2 A method of operation according to claim 1, wherein the step of determining the knowledge item to be diagnosed comprises: determining, by the knowledge state inference device, the number of times each of the plurality of knowledge items is included in the residual knowledge states; and determining by the knowledge state inference device the knowledge item having the smallest number of times among the counts that is the smallest difference from half the number of the residual knowledge states. Claim 3 A method of operation according to claim 1, wherein the step of transmitting the item comprises: a step of loading at least one item among a plurality of items corresponding to the knowledge item to be diagnosed from the item database of the server by the knowledge state inference device; and a step of transmitting the loaded item to the user device by the knowledge state inference device. Claim 4 A method of operation comprising: determining the number of remaining knowledge states after some of the remaining knowledge states have been erased based on the user response by the knowledge state inference device; repeating the steps of determining the knowledge item to be diagnosed, transmitting the item, receiving the user response, and erasing some of the remaining knowledge states after some have been erased, in response to the number of the remaining knowledge states being 2 or more by the knowledge state inference device; and determining the remaining knowledge state as the current knowledge state in response to the number of the remaining knowledge states being 1 by the knowledge state inference device. Claim 5 In claim 1, the at least one learning unit comprises at least one combination of a single sub-unit and a plurality of sub-units, and the plurality of sub-units are a method of operation included in at least some of a subject, a major unit, different subjects, and different major units. Claim 6 In claim 1, the plurality of knowledge items include diagnostic knowledge items and learning knowledge items, and the plurality of vectors include first vectors and second vectors, and the step of loading the plurality of knowledge items and the plurality of vectors respectively comprises: loading the learning knowledge items included in the at least one learning unit from the server by the knowledge state inference device; loading the first vectors pointing to the learning knowledge items as upper knowledge items from the server by the knowledge state inference device; determining the knowledge items pointed to by the first vectors as lower knowledge items as the diagnostic knowledge items by the knowledge state inference device; and loading the second vectors pointing to the diagnostic knowledge items as upper knowledge items from the server by the knowledge state inference device, and the step of generating the knowledge space comprises: generating a learning knowledge space including learning knowledge states having the diagnostic knowledge items and the learning knowledge items as elements based on the diagnostic knowledge items, the learning knowledge items, the first vectors, and the second vectors by the knowledge state inference device; A method of operation comprising the step of generating a diagnostic knowledge space including diagnostic knowledge states having diagnostic knowledge items as elements among the learning knowledge states by the knowledge state inference device, and the step of determining the current knowledge state including the step of determining the current knowledge state among the diagnostic knowledge states included in the diagnostic knowledge space by the knowledge state inference device. Claim 7 In claim 6, the step of determining the target knowledge item comprises: determining, by the curriculum recommendation device, at least one higher knowledge state having the current knowledge state as a subset among the learning knowledge states included in the learning knowledge space; performing a difference operation between the at least one higher knowledge state and the current knowledge state by the curriculum recommendation device; and determining, by the curriculum recommendation device, at least one knowledge item that can be added to the current knowledge state as the target knowledge item based on the result of the difference operation, and the step of outputting the system response comprises, by the curriculum recommendation device, outputting a system response requesting the user device to select at least one of the target knowledge items. Claim 8 A method of operation according to claim 7, further comprising: receiving a second user input from the user device selecting a first target knowledge item among the target knowledge items by the curriculum recommendation device; transmitting a learning curriculum to the user device for learning the first target knowledge item by the curriculum recommendation device; transmitting an additional question to the user device for diagnosing whether the first target knowledge item has been learned by the knowledge state inference device; receiving an additional user response to the additional question from the user device by the knowledge state inference device; and updating the current knowledge state based on the additional user response by the knowledge state inference device. Claim 9 A method of operation according to claim 8, further comprising: a step of determining by the knowledge state inference device whether the updated current knowledge state is identical to the final knowledge state; and a step of transmitting a plurality of final questions to the user device for diagnosing knowledge items included in the final knowledge state in response to the knowledge state inference device determining that the updated current knowledge state is identical to the final knowledge state, wherein the final knowledge state is a knowledge state that includes the most learning knowledge items among the learning knowledge states included in the learning knowledge space. Claim 10 A Knowledge Space Theory (KST)-based adaptive learning system communicating with a user device comprises: a server including a first database storing multiple knowledge items and a second database storing multiple vectors representing hierarchical relationships between the multiple knowledge items; receiving a first user input pointing to at least one learning unit from the user device, loading the multiple knowledge items and the multiple vectors corresponding to the at least one learning unit from the server, generating sets having the multiple knowledge items as elements as multiple candidate knowledge states, determining a candidate knowledge state that includes a higher-level knowledge item and does not include a lower-level knowledge item of the higher-level knowledge item based on the multiple vectors as an error knowledge state, and by eliminating the error knowledge state, generating a knowledge space containing multiple knowledge states, setting the multiple knowledge states included in the knowledge space as residual knowledge states, determining the knowledge item with the greatest branching effect on the residual knowledge states as a knowledge item to be diagnosed, transmitting a question corresponding to the knowledge item to be diagnosed to the user device, receiving a user response to the question from the user device, and in response to the user response being correct, the A knowledge state inference device configured to eliminate knowledge states among the remaining knowledge states that do not include the knowledge item to be diagnosed, eliminate knowledge states among the remaining knowledge states that include the knowledge item to be diagnosed in response to the user's response being incorrect, and determine the user's current knowledge state based on the remaining knowledge states after elimination according to the user's response;A KST-based adaptive learning system comprising a curriculum recommendation device configured to determine at least one target knowledge item that can be added to the current knowledge state based on the current knowledge state and to output a system response recommending a learning curriculum for learning the target knowledge item to the user device, wherein the diagnosis target knowledge item is selected such that the number of erasable knowledge states among the remaining knowledge states is maximized based on the user response to the diagnosis target knowledge item.

Images