Intelligent teaching system interoperation model based on subject knowledge graph

Abstract

The invention provides an intelligent teaching system interoperation model based on a subject knowledge graph through expansion of SCORM specifications, and the intelligent teaching system interoperation model comprises the following sub-models: a sub-model 1, a structural framework of the subject knowledge graph; the sub-model 2 is a sharable intelligent content object KCO based on a subject knowledge graph; a sub-model 3 is a KCO operation environment conforming to the SCORM specification; and a sub-model 4 is a KCO content packaging model KCOCP conforming to the SCORM specification. According to the invention, the subject knowledge graph is used as a knowledge base, and the content model of the SCORM specification is expanded from structured courseware based on multimedia to intelligent learning content fused with the subject knowledge graph, so that the overall intelligence of a teaching system conforming to the SCORM specification can be greatly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of the intersection of artificial intelligence and educational technology, and specifically relates to an interoperability method of an intelligent teaching system based on a subject knowledge graph. Background Art

[0002] In recent years, with the continuous launch of Google knowledge graph related products, the knowledge graph technology has also attracted widespread attention from the industry and academia. It has become one of the key technologies for the transformation and upgrading of various industries from networking to intelligence. It also provides a new technical means for the core problem of domain knowledge modeling in intelligent teaching systems. Therefore, building a knowledge graph in the field of education has become an important research topic in the development of intelligent teaching systems. At present, the most typical intelligent teaching system technical specification in the world is the SCORM specification [1] published by the US federal government organization ADL and incorporated into the ISO standard system. This specification has been supported by many software platforms including Adobe. At the same time, based on this standard, China has carried out relevant research on learning cells [2] and fragmented [3] learning proposed by Professor Yu Shengquan of Beijing Normal University. Zhang Shuqin et al. developed a technical architecture of university smart learning environment that complies with the SCORM specification [4], and Lin Longcheng studied the combination of the new e-learning standard xAPI and the SCORM specification [5]. However, the domain knowledge model of the SCORM specification only uses structured courseware for students, and does not introduce a subject expert model based on the domain knowledge graph. It only implements an adaptive teaching expert system that can provide learners with personalized teaching based on structured courseware and production rules, and its overall intelligence is not high. Therefore, how to integrate the subject knowledge graph into the SCORM specification and build a more efficient and intelligent intelligent teaching system interoperability model has become one of the key technologies to accelerate the application of educational knowledge graphs in intelligent teaching systems.

[0003] Based on the SCO sharable content object of ADL, the present invention proposes a sharable and automatically reasonable intelligent content object KCO based on the subject knowledge graph. KCO is a special SCO content object that can communicate with the learning management system (LMS) during operation and access the subject knowledge graph and its intelligent teaching functions in a unified way, so that it can be moved and interoperated between heterogeneous learning management systems. Then, by extending the ADL / SCORM specification, an interoperability model KCOIM of the KCO intelligent content object is formed, including: (1) using the proposed subject knowledge graph general model as the subject knowledge model in the SCORM specification; (2) extending the API method in the SCORM specification running environment to enable it to query the subject knowledge graph and its intelligent functions; (3) adding a knowledge graph CKG and its associated learner cognitive state LCS two-level data structure in the communication data model of the SCORM specification running environment to encapsulate the design details of the subject knowledge graph and its learner model; (4) adding a subject graph node in the content aggregation model of the SCORM specification so that the same intelligent content object KCO can be associated with different subject knowledge graphs. Ultimately, the content model of the SCORM specification will be expanded from multimedia-based structured courseware to intelligent learning content that integrates subject knowledge graphs, providing the possibility for interoperable knowledge graph-based learning content and learning sequences, and greatly improving the overall intelligence of teaching systems that comply with the SCORM specification.

[0004] References

[0005] [1] ADL. SCORM Versions and Resources [EB / OL]. 2024.12.11. Available at: https: / / adlnet.gov / past-projects / scorm / #scorm-versions-and-resources.

[0006] [2] Yu Shengquan, Duan Jinju, Cui Jingjing. Double helix deep learning model based on learning element[J]. Modern Distance Education Research, 2017(06):37-47+56.

[0007] [3] Yang Zekai, Zhou Haijun. A review of research on fragmented learning in the new media era. Digital Media Research, 2018, 35(7): 20-27.

[0008] [4] Zhang Shuqin, Wang Jinyang, Bai Guangyao, Zhang Minzhi. Research on the technical architecture of smart learning environment in universities. China Education Informatization, 2021, 2021.21:62-66.

[0009] [5] Lin Longcheng. Research on the application of the new generation of learning behavior description specification xAPI. Information Technology and Informatization, 2021, 2021.5:230-232. Summary of the invention

[0010] The present invention discloses an intelligent teaching system interoperability model based on subject knowledge graph, which is characterized by comprising the following sub-models:

[0011] Sub-model 1. Structural framework of subject knowledge graph: used to support intelligent teaching. The framework specifically includes: (1) concept schema, (2) content schema, (3) knowledge schema, (4) cognitive schema, and (5) structural definition of relationships;

[0012] Sub-model 2. Sharable intelligent content object KCO based on subject knowledge graph: It is a special SCO sharable content object in the SCORM specification. It can communicate with the learning management system LMS during operation. At the same time, KCO is extended to access the subject knowledge graph and its intelligent teaching function in a unified way, so as to design adaptive learning content according to the learner's ability and performance;

[0013] Sub-model 3. KCO runtime environment that complies with the SCORM specification: including a knowledge graph communication data structure that encapsulates the design details of the subject knowledge graph, a learner cognitive state communication data structure that encapsulates the design details of the learner model, and a set of distributed query communication interfaces of the subject knowledge graph, which are used by KCO objects to access the subject knowledge graph in a unified manner;

[0014] Sub-model 4. KCO content packaging model KCOCP that complies with the SCORM specification: used to define the interoperable content structure based on KCO objects, using the extensible markup language XML to describe the content manifest file that complies with the SCORM specification, requiring KCO not to specify the subject knowledge graph to be operated during the design phase, and deferring the association between KCO and the subject knowledge graph it operates in the content packaging manifest until runtime;

[0015] The SCORM specification is a Sharable Content Object Reference Model proposed by the Advanced Distributed Learning Group (ADL), an organization of the US federal government.

[0016] Furthermore, the specific features of the sub-model 1 are:

[0017] (1) The concept schema is used to reflect the concepts, concept synonyms and concept classification structure of the subject area. It is composed of the concept synonym set and the "is-a" classification diagram. It is a constraint model of the subject knowledge graph and can provide concept associations for the content schema, knowledge schema and cognitive schema.

[0018] (2) The content schema is used to reflect multimedia teaching content, which is composed of media relationships. The media targets include documents, pictures, animations, videos and multimedia teaching content units of KCO objects. The knowledge of a certain aspect of the concept is explained in the form of documents, animations and videos with mixed text and pictures;

[0019] (3) The knowledge schema is used to reflect the subject knowledge of concepts, and is composed of semantic relations and attribute relations of concepts, wherein the semantic relations of concepts are used to reflect the connection between concepts in the subject field, and the attribute relations are used to reflect the connection between concepts and attribute entities, wherein the attribute entities refer to data reflecting the characteristics of concepts, including text, time and numerical values;

[0020] (4) The cognitive schema is used to reflect the learning order relationship between concepts, which is composed of "predecessor" relationship and "successor" relationship;

[0021] (5) The structural definition of the relationship is characterized in that the relationship in the subject knowledge graph is defined as a four-tuple separated by a slash “ / ” as follows:

[0022] <relationship>:=<subject name> / <concept set> / <relationship name> / <target type> (1)

[0023] Among them, "discipline name" indicates the subject field to which the relationship belongs; "concept set" is used to give the definition domain of the relationship, indicating the scope of concepts with defined relationships. If all concepts have defined relationships, "ontology" is used to represent the concept set of the relationship; "relationship name" indicates the name of the relationship, which is named using verbs, nouns and noun phrases; "target type" is used to give the value domain of the relationship, indicating the type of knowledge to which the relationship points, and the values ​​are five target types: "concept", "attribute", "static content", "dynamic content" and "KCO object". Among them, "concept" means that the relationship points to the concept in the concept schema, "attribute" means that the relationship points to the data entity of text, time and value, "static content" means that the relationship points to the web-type teaching content mixed with text and pictures, "dynamic content" means that the relationship points to the teaching content of animation and video files, and "KCO object" means that the relationship points to a shareable intelligent content object.

[0024] Furthermore, the specific structure of the sub-model 2 is: a KCO content object at least includes an API adapter object for locating LMS, an initialization API call Initialize() for establishing a dialogue with the teaching system and a termination API call Terminate() for ending the dialogue, as well as API calls GetValue() and SetValue() for accessing subject knowledge graphs, where API is the abbreviation of application program interface.

[0025] Furthermore, the sub-model 3 specifically includes:

[0026] Sub-model 3.1 Knowledge graph communication data structure cgg, including communication elements representing the hierarchical structure shown in 1:

[0027] Table 1 Hierarchical data structure of knowledge graph ckg

[0028] level Data elements in cmi.ckg describe Access Rights 1 cmi.ckg.concept Query Concept W+R 1.1 cmi.ckg.concept.attributeSet The concept's attribute set R 1.2 cmi.ckg.concept.attribute Attributes of concepts W+R 1.2.1 cmi.ckg.concept.attribute.hasValue Concept attribute value R 1.3 cmi.ckg.concept.relationSet Conceptual set of relationships R 1.4 cmi.ckg.concept.relation Conceptual Relationship W+R 1.4.1 cmi.ckg.concept.relation.target Conceptual Relationship Target R 1.5 cmi.ckg.concept.instanceSet A collection of instances of a concept R 1.6 cmi.ckg.concept.synonymSet Synonym set for concept R 1.7 cmi.ckg.concept.parents Direct superiority of concept R 1.8 cmi.ckg.concept.directHyponymSet The immediate subordinate set of a concept R 1.9 cmi.ckg.concept.relation.disturbSet Interference set of concept relations R 1.10 cmi.ckg.concept.nextConcept Next learning concept R 2 cmi.ckg.root The root node of the is-a classification structure R 2.1 cmi.ckg.root.allHyponymNum All the lower digits of the root node R 3 cmi.ckg.KnowledgeGraph Knowledge Graph R 3.1 cmi.ckg.KnowledgeGraph.discipline Knowledge Graph Disciplines R

[0029] Sub-model 3.2 The communication data structure lcs of the learner's cognitive state includes the communication elements of the hierarchical structure shown in Table 2:

[0030] Table 2 Hierarchical data structure of learner cognitive state LCS

[0031]

[0032]

[0033] Sub-model 3.3 Distributed query communication interface of subject knowledge graph: extends the functions of the six API methods in the SCORM specification, including:

[0034] (1) Initialize(): It is extended to activate the learning system LMS to initialize the communication dialogue with the KCO object, including creating a knowledge graph communication data structure cgg and a learner cognitive state communication data structure lcs for the communicating KCO object on the server side. Its interface format is: return_value = Initialize(parameter), where parameter parameter is an empty string, and return value return_value is a logical value "true" or "false", indicating whether the initialization communication dialogue is successful;

[0035] (2) Terminate(): It is extended to terminate the communication dialogue of a KCO object, including releasing the knowledge graph communication data structure cgg and the learner cognitive state communication data structure lcs established for the KCO object. Its interface format is: return_value = Terminate(parameter), where parameter parameter is an empty string, and the return value return_value is a logical value "true" or "false", indicating whether the communication dialogue is terminated successfully;

[0036] (3) SetValue(): It is extended to enable the KCO object to set the communication data elements in ckg and lcs. Its interface format is: return_value = SetValue (parameter1, parameter2), where parameter1 is the first string parameter, indicating the communication data element of ckg and lcs to be set, parameter2 is the second string parameter, indicating the value of the communication data element specified by parameter1, and the return value return_value is a logical value "true" or "false", indicating whether the setting is successful;

[0037] (4) GetValue(): extended to be able to pass the value of the communication data element specified in the ckg and lcs on the server side to the KCO object on the client side as the query result. Its interface format is: return_value = GetValue(parameter), where the parameter parameter represents the communication data element to be queried, and the return value return_value is the query result;

[0038] (5) GetLastError(): extended to return the error code of the most recent access to ckg and lcs. Its interface format is: return_value = GetError(), where the return value symbol return_value receives the returned error code number;

[0039] (6) GetErrorString(): extended to be able to return the error information corresponding to the error code of the most recent access to ckg and lcs. Its interface format is: return_value = GetErrorString(parameter), parameter parameter is the error code number, and the return value symbol return_value receives the returned error information string.

[0040] Furthermore, the learner model described in sub-model 3 is based on the concept schema in the subject knowledge graph, and the concepts in the concept schema are used as assessment points of the subject knowledge in the learner model.

[0041] Furthermore, the sub-model 4 requires that KCO does not specify the subject knowledge graph to be operated in the design phase, and postpones the association between KCO and the subject knowledge graph it operates in the content packaging list until runtime. The specific implementation method is as follows;

[0042] (1) Expand a knowledge graph element in the SCORM content packaging model <knowledge-graph>, used to describe the subject knowledge graph used in the content package;

[0043] (2) Project elements in the SCORM content packaging model <item>In , an optional attribute knowledgeref of knowledge graph reference is extended to describe the association between the KCO object and the subject knowledge graph it references;

[0044] (3) Resource elements in the SCORM content packaging model <resource>In, extend an optional attribute kcoimType describing the KCO interoperability model type and the corresponding attribute value "KCO";

[0045] (4) Extended <knowledge-graph> 、 <item>and <resource>The elements are defined in the XML schema file kcocp.xsd.

[0046] Furthermore, the extended knowledge graph element <knowledge-graph>The XML format is as follows:

[0047] (1) Description: <knowledge-graph>Elements describe the subject knowledge graph used in content packaging;

[0048] (2) Diversity: <knowledge-graph>Element in the manifest element <manifest>It may appear 0 or 1 times;

[0049] (3) Attribute identifier: A unique identifier in the content manifest, which must be specified;

[0050] (4) Attribute discipline: a string describing the discipline to which the discipline knowledge graph belongs, which must be specified;

[0051] (5) Attribute href: a reference to the entry file of the subject knowledge graph, which must be specified;

[0052] (6) Sub-elements <metadata>: Metadata describing the subject knowledge graph, optional element;

[0053] (7) Sub-elements <file>: A list of files that the subject knowledge graph depends on, which must be specified.

[0054] The present invention has the following characteristics and advantages:

[0055] (1) By expanding the ADL / SCORM specification, an interoperability model of intelligent teaching systems based on subject knowledge graphs is formed;

[0056] (2) With the support of the interoperability model proposed in this invention, the proposed general model of subject knowledge graph can be used as the subject knowledge model in the SCORM specification;

[0057] (3) Expanded the API methods in the SCORM specification operating environment to enable querying of subject knowledge graphs and their intelligent functions;

[0058] (4) In the communication data model of the SCORM standard operating environment, a knowledge graph CKG and its associated learner cognitive state LCS are added to realize the encapsulation of the design details of the subject knowledge graph and its learner model;

[0059] (5) Expand the content model of the SCORM specification from multimedia-based structured courseware to intelligent learning content that integrates subject knowledge graphs, providing the possibility for interoperable learning content and learning sequences based on knowledge graphs, and greatly improving the overall intelligence of teaching systems that comply with the SCORM specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is the runtime environment of the interoperability model of intelligent teaching systems based on subject knowledge graphs;

[0061] Figure 2 It is the structural model of the subject knowledge graph;

[0062] Figure 3 It is a shareable intelligent content object KCO framework structure based on subject knowledge graph;

[0063] Figure 4 It is a schematic diagram of the expanded SCORM / SCOCP content packaging structure. Specific implementation methods

[0065] The present invention is further described below in conjunction with specific embodiments and drawings, but the protection scope of the present invention is not limited to the following embodiments.

[0066] An interoperability model of intelligent teaching system based on subject knowledge graph, the runtime environment is as follows Figure 1 As shown, it is characterized by including the following sub-models:

[0067] Sub-model 1. Structural framework of subject knowledge graph: used to support intelligent teaching. The framework specifically includes: (1) concept schema, (2) content schema, (3) knowledge schema, (4) cognitive schema, and (5) structural definition of relationships;

[0068] Sub-model 2. Sharable intelligent content object KCO based on subject knowledge graph: It is a special SCO sharable content object in the SCORM specification. It can communicate with the learning management system LMS during operation. At the same time, KCO is extended to access the subject knowledge graph and its intelligent teaching function in a unified way, so as to design adaptive learning content according to the learner's ability and performance;

[0069] Sub-model 3. KCO runtime environment that complies with the SCORM specification: including a knowledge graph communication data structure that encapsulates the design details of the subject knowledge graph, a learner cognitive state communication data structure that encapsulates the design details of the learner model, and a set of distributed query communication interfaces of the subject knowledge graph, which are used by KCO objects to access the subject knowledge graph in a unified manner;

[0070] Sub-model 4. KCO content packaging model KCOCP that complies with the SCORM specification: used to define the interoperable content structure based on KCO objects, using the extensible markup language XML to describe the content manifest file that complies with the SCORM specification, requiring KCO not to specify the subject knowledge graph to be operated during the design phase, and deferring the association between KCO and the subject knowledge graph it operates in the content packaging manifest until runtime;

[0071] The SCORM specification is a Sharable Content Object Reference Model proposed by the Advanced Distributed Learning Group (ADL), an organization of the US federal government.

[0072] Furthermore, the structural model of the sub-model 1 is as follows Figure 2 As shown, the specific features are:

[0073] (1) The concept schema is used to reflect the concepts, concept synonyms and concept classification structure of the subject area. It is composed of the concept synonym set and the "is-a" classification diagram. It is a constraint model of the subject knowledge graph and can provide concept associations for the content schema, knowledge schema and cognitive schema.

[0074] (2) The content schema is used to reflect multimedia teaching content, which is composed of media relationships. The media targets include documents, pictures, animations, videos and multimedia teaching content units of KCO objects. The knowledge of a certain aspect of the concept is explained in the form of documents, animations and videos with mixed text and pictures;

[0075] (3) The knowledge schema is used to reflect the subject knowledge of concepts, and is composed of semantic relations and attribute relations of concepts, wherein the semantic relations of concepts are used to reflect the connection between concepts in the subject field, and the attribute relations are used to reflect the connection between concepts and attribute entities, wherein the attribute entities refer to data reflecting the characteristics of concepts, including text, time and numerical values;

[0076] (4) The cognitive schema is used to reflect the learning order relationship between concepts, which is composed of "predecessor" relationship and "successor" relationship;

[0077] (5) The structural definition of the relationship is characterized in that the relationship in the subject knowledge graph is defined as a four-tuple separated by a slash " / ":

[0078] <relationship>:=<subject name> / <concept set> / <relationship name> / <target type> (1)

[0079] Among them, "discipline name" indicates the discipline field to which the relationship belongs; "concept set" is used to give the definition domain of the relationship, indicating the scope of concepts with defined relationships. If all concepts have defined relationships, "ontology" is used to represent the concept set of the relationship; "relationship name" indicates the name of the relationship, which is named using verbs, nouns and noun phrases; "target type" is used to give the value domain of the relationship, indicating the type of knowledge the relationship points to, and its values ​​are five target types: "concept", "attribute", "static content", "dynamic content" and "KCO object". Among them, "concept" indicates that the relationship points to the concept in the concept schema, "attribute" indicates that the relationship points to the data entity of text, time and value, "static content" indicates that the relationship points to the webpage teaching content mixed with text and pictures, "dynamic content" indicates that the relationship points to the teaching content of animation and video files, and "KCO object" indicates that the relationship points to a shareable intelligent content object.

[0080] Furthermore, the specific model structure of the sub-model 2 is as follows: a KCO content object at least includes an API adapter object for locating the LMS, an initialization API call Initialize() for establishing a dialogue with the teaching system and a termination API call Terminate() for ending the dialogue, as well as API calls GetValue() and SetValue() for accessing the subject knowledge graph, such as Figure 3 shown.

[0081] Furthermore, the sub-model 3 specifically includes:

[0082] Sub-model 3.1 Knowledge graph communication data structure cgg, including communication elements representing the hierarchical structure shown in 1:

[0083] Table 1 Hierarchical data structure of knowledge graph ckg

[0084] level Data elements in cmi.ckg describe Access Rights 1 cmi.ckg.concept Query Concept W+R 1.1 cmi.ckg.concept.attributeSet The concept's attribute set R 1.2 cmi.ckg.concept.attribute Attributes of concepts W+R 1.2.1 cmi.ckg.concept.attribute.hasValue Concept attribute value R 1.3 cmi.ckg.concept.relationSet Conceptual set of relationships R 1.4 cmi.ckg.concept.relation Conceptual Relationship W+R 1.4.1 cmi.ckg.concept.relation.target Conceptual Relationship Target R 1.5 cmi.ckg.concept.instanceSet A collection of instances of a concept R 1.6 cmi.ckg.concept.synonymSet Synonym set for concept R 1.7 cmi.ckg.concept.parents Direct superiority of concept R 1.8 cmi.ckg.concept.directHyponymSet The immediate subordinate set of a concept R 1.9 cmi.ckg.concept.relation.disturbSet Interference set of concept relations R 1.10 cmi.ckg.concept.nextConcept Next learning concept R 2 cmi.ckg.root The root node of the is-a classification structure R 2.1 cmi.ckg.root.allHyponymNum All the lower digits of the root node R 3 cmi.ckg.KnowledgeGraph Knowledge Graph R 3.1 cmi.ckg.KnowledgeGraph.discipline Knowledge Graph Disciplines R

[0085] Sub-model 3.2 The communication data structure lcs of the learner's cognitive state includes the communication elements of the hierarchical structure shown in Table 2:

[0086] Table 2 Hierarchical data structure of learner cognitive state LCS

[0087]

[0088]

[0089] Sub-model 3.3 Distributed query communication interface of subject knowledge graph: extends the functions of the six API methods in the SCORM specification, including:

[0090] (1) Initialize(): It is extended to activate the learning system LMS to initialize the communication dialogue with the KCO object, including creating a knowledge graph communication data structure cgg and a learner cognitive state communication data structure lcs for the communicating KCO object on the server side. Its interface format is: return_value = Initialize(parameter), where parameter parameter is an empty string, and return value return_value is a logical value "true" or "false", indicating whether the initialization communication dialogue is successful;

[0091] (2) Terminate(): It is extended to terminate the communication dialogue of a KCO object, including releasing the knowledge graph communication data structure cgg and the learner cognitive state communication data structure lcs established for the KCO object. Its interface format is: return_value = Terminate(parameter), where parameter parameter is an empty string, and the return value return_value is a logical value "true" or "false", indicating whether the communication dialogue is terminated successfully;

[0092] (3) SetValue(): It is extended to enable the KCO object to set the communication data elements in ckg and lcs. Its interface format is: return_value = SetValue (parameter1, parameter2), where parameter1 is the first string parameter, indicating the ckg and lcs communication data element to be set, parameter2 is the second string parameter, indicating the value of the communication data element specified by parameter1, and the return value return_value is a logical value "true" or "false", indicating whether the setting is successful;

[0093] (4) GetValue(): extended to be able to pass the value of the communication data element specified in the ckg and lcs on the server side to the KCO object on the client side as the query result. Its interface format is: return_value = GetValue(parameter1, parameter2), where parameter parameter1 represents the communication data element to be queried, parameter parameter2 is the name of a variable in the KCO object that receives the query result, and the return value return_value is a logical value "true" or "false", indicating whether the query is successful;

[0094] (5) GetLastError(): extended to return the error code of the most recent access to ckg and lcs. Its interface format is: return_value = GetError(), where the return value symbol return_value receives the returned error code number;

[0095] (6) GetErrorString(): extended to be able to return the error information corresponding to the error code of the most recent access to ckg and lcs. Its interface format is: return_value = GetErrorString(parameter), parameter parameter is the error code number, and the return value symbol return_value receives the returned error information string.

[0096] Furthermore, the learner model described in sub-model 3 is based on the concept schema in the subject knowledge graph, and the concepts in the concept schema are used as assessment points of the subject knowledge in the learner model.

[0097] Furthermore, the sub-model 4 requires that KCO does not specify the subject knowledge graph to be operated in the design phase, and postpones the association between KCO and the subject knowledge graph it operates in the content packaging list until runtime. The specific implementation method is as follows;

[0098] (1) Expand a knowledge graph element in the SCORM content packaging model <knowledge-graph>, used to describe the subject knowledge graph used in the content package;

[0099] (2) Project elements in the SCORM content packaging model <item>In , an optional attribute knowledgeref of knowledge graph reference is extended to describe the association between the KCO object and the subject knowledge graph it references;

[0100] (3) Resource elements in the SCORM content packaging model <resource>In, extend an optional attribute kcoimType describing the KCO interoperability model type and the corresponding attribute value "KCO";

[0101] (4) Extended <knowledge-graph> 、 <item>and <resource>The elements are defined in the XML schema file kcocp.xsd.

[0102] Furthermore, the extended knowledge graph element <knowledge-graph>The XML format is as follows:

[0103] (1) Description: <knowledge-graph>Elements describe the subject knowledge graph used in content packaging;

[0104] (2) Diversity: <knowledge-graph>Element in the manifest element <manifest>It may appear 0 or 1 times;

[0105] (3) Attribute identifier: A unique identifier in the content manifest, which must be specified;

[0106] (4) Attribute discipline: a string describing the discipline to which the discipline knowledge graph belongs, which must be specified;

[0107] (5) Attribute href: a reference to the entry file of the subject knowledge graph, which must be specified;

[0108] (6) Sub-elements <metadata>: Metadata describing the subject knowledge graph, optional element;

[0109] (7) Sub-elements <file>: A list of files that the subject knowledge graph depends on, which must be specified.

[0110] Application Examples

[0111] In the KCOIM content packaging model KCOCP of this application example, the extended <knowledge-graph>It is used to describe the subject knowledge graph used in the content package. Its XML format is as follows:

[0112] <knowledge-graph>element

[0113] Description: This element describes the subject knowledge graph used in the content package.

[0114] Diversity: This element is <manifest>May appear 0 or 1 times in an element.

[0115] property:

[0116] identifier (required). A unique identifier in the manifest.

[0117] discipline (required). A string describing the discipline to which the discipline knowledge graph belongs.

[0118] href (required). A reference to the entry file of the subject knowledge graph.

[0119] Child elements:

[0120] · <metadata>(Optional): Metadata describing the subject knowledge graph.

[0121] · <file>(Optional): A list of files that the subject knowledge graph depends on. This element needs to be repeated for each file in a given ontology. <file>The element refers to a file local to the content package through the href attribute.

[0122] In the KCOIM content packaging model KCOCP of this application example, the association between the KCO object and the subject knowledge graph it references is in the content list <organization>Hierarchy sub-elements <item>To achieve this association, this example is the content packaging model of the SCORM specification. <item>The element adds an optional knowledgeref attribute. Additionally, to identify our proposed KCO content object in the "Resource" section of the manifest, we define a knowledgeref attribute for <resource>The element is extended with an attribute kcoimType and a corresponding attribute value "KCO". Finally, all extended elements are defined in the XML schema file kcocp.xsd, which contains an assumed schema namespace: "http: / / www.kcoim.org / xsd / icocp_v1p0", thereby extending the structure of the SCORM content package SCOCP from a three-layer structure to a four-layer structure, such as Figure 4 shown.

[0123] In this application example, it is assumed that an intelligent question-answering system based on a subject knowledge graph has been designed. The web homepage of the question-answering system is Answer.html. Then, the KCO content packaging model designed by us is applied to identify the intelligent question-answering system as a Q&A KCO object, and associate it with a computer subject knowledge graph consisting of three files: kg1.date, kg2.date, and kg3.date. The corresponding content packaging list file is as follows:

[0124] < / resource> < / item> < / item> < / organization> < / file> < / file> < / metadata> < / manifest> < / file> < / metadata> < / manifest> < / resource> < / item> < / knowledge-graph> < / resource> < / item> < / file> < / metadata> < / manifest> < / resource> < / item> < / knowledge-graph> < / resource> < / item>

Claims

1. The interoperability model of intelligent teaching system based on subject knowledge graph is characterized by Includes the following sub-models: Sub-model 1. Structural framework of subject knowledge graph: used to support intelligent teaching. The framework specifically includes: (1) concept schema, (2) content schema, (3) knowledge schema, (4) cognitive schema, and (5) structural definition of relationships; Sub-model 2. Sharable intelligent content object KCO based on subject knowledge graph: It is a special SCO sharable content object in the SCORM specification. It can communicate with the learning management system LMS during operation. At the same time, KCO is extended to access the subject knowledge graph and its intelligent teaching function in a unified way, so as to design adaptive learning content according to the learner's ability and performance; Sub-model 3. KCO runtime environment that complies with the SCORM specification: including a knowledge graph communication data structure that encapsulates the design details of the subject knowledge graph, a learner cognitive state communication data structure that encapsulates the design details of the learner model, and a set of distributed query communication interfaces of the subject knowledge graph, which are used by KCO objects to access the subject knowledge graph in a unified manner; Sub-model 4. KCO content packaging model KCOCP that complies with the SCORM specification: used to define the interoperable content structure based on KCO objects, using the extensible markup language XML to describe the content manifest file that complies with the SCORM specification, requiring KCO not to specify the subject knowledge graph to be operated during the design phase, and deferring the association between KCO and the subject knowledge graph it operates in the content packaging manifest until runtime; The SCORM specification is a Sharable Content Object Reference Model proposed by the Advanced Distributed Learning Group (ADL), an organization of the US federal government.

2. The interoperability model of the intelligent teaching system based on subject knowledge graph according to claim 1 is characterized by: The specific features of the sub-model 1 are: (1) The concept schema is used to reflect the concepts, concept synonyms and concept classification structure of the subject area. It is composed of the concept synonym set and the "is-a" classification diagram. It is a constraint model of the subject knowledge graph and can provide concept associations for the content schema, knowledge schema and cognitive schema. (2) The content schema is used to reflect multimedia teaching content, which is composed of media relationships. The media targets include documents, pictures, animations, videos and multimedia teaching content units of KCO objects. The knowledge of a certain aspect of the concept is explained in the form of documents, animations and videos with mixed text and pictures; (3) The knowledge schema is used to reflect the subject knowledge of concepts, and is composed of semantic relations and attribute relations of concepts, wherein the semantic relations of concepts are used to reflect the connection between concepts in the subject field, and the attribute relations are used to reflect the connection between concepts and attribute entities, wherein the attribute entities refer to data reflecting the characteristics of concepts, including text, time and numerical values; (4) The cognitive schema is used to reflect the learning order relationship between concepts, which is composed of "predecessor" relationship and "successor" relationship; (5) The structural definition of the relationship is characterized in that the relationship in the subject knowledge graph is defined as a four-tuple separated by a slash " / ": <relationship>:=<subject name> / <concept set> / <relationship name> / <target type> (1) Among them, "discipline name" indicates the discipline field to which the relationship belongs; "concept set" is used to give the definition domain of the relationship, indicating the scope of concepts with defined relationships. If all concepts have defined relationships, "ontology" is used to represent the concept set of the relationship; "relationship name" indicates the name of the relationship, which is named using verbs, nouns and noun phrases; "target type" is used to give the value domain of the relationship, indicating the type of knowledge the relationship points to, and its values ​​are five target types: "concept", "attribute", "static content", "dynamic content" and "KCO object". Among them, "concept" indicates that the relationship points to the concept in the concept schema, "attribute" indicates that the relationship points to the data entity of text, time and value, "static content" indicates that the relationship points to the webpage teaching content mixed with text and pictures, "dynamic content" indicates that the relationship points to the teaching content of animation and video files, and "KCO object" indicates that the relationship points to a shareable intelligent content object.

3. The interoperability model of the intelligent teaching system based on subject knowledge graph according to claim 2 is characterized by: The specific structure of the sub-model 2 is: a KCO content object at least includes an API adapter object for locating LMS, an initialization API call Initialize() for establishing a dialogue with the teaching system and a termination API call Terminate() for ending the dialogue, as well as API calls GetValue() and SetValue() for accessing the subject knowledge graph. The API is the abbreviation of application program interface.

4. The interoperability model of the intelligent teaching system based on subject knowledge graph according to claim 3 is characterized by: The sub-model 3 specifically includes: Sub-model 3.1 Knowledge graph communication data structure ckg: includes the following hierarchical communication elements: (1) Query concept element cmi.ckg.concept: represents the concept currently queried by the KCO object, access rights: W+R, where: the symbol "cmi" represents the abbreviation of Computer Managed Instruction, "ckg" represents the knowledge graph communication data structure, "concept" represents the concept, "W" represents the write permission, which means that the KCO object can set the value of the element, "R" represents the read permission, which means that the KCO object can query the value of the element, and ". " represents the hierarchical structure; (2) Concept attribute set element cmi.ckg.concept.attributeSet: represents the attribute set of the current query concept, access permission: R, where the symbol "attributeSet" represents the attribute set; (3) Concept attribute element cmi.ckg.concept.attribute: represents the current query attribute of the current query concept, access rights: W+R, where: the symbol "attribute" represents the attribute; (4) Concept attribute value element cmi.ckg.concept.attribute.hasValu: indicates the value of the current query attribute, access right: R, where: the symbol "hasValue" indicates the value; (5) Concept relation set element cmi.ckg.concept.relationSet: represents the relation set of the current query concept, access permission: R, where the symbol "relationSet" represents the relation set; (6) Concept relation element cmi.ckg.concept.relation: represents the current query relation of the current query concept, access rights: W+R, where the symbol "relation" represents a relation; (7) Conceptual relation target element cmi.ckg.concept.relation.target: represents the target of the current query relation, access right R, where: the symbol "target" represents the target; (8) Concept instance set element cmi.ckg.concept.instanceSet: represents the instance set of the current query concept, access right: R, where: the symbol "instanceSet" represents the instance set; (9) Concept synonym set element cmi.ckg.concept.synonymSet: represents the synonym set of the current query concept, access right: R, where: the symbol "synonymSet" represents the synonym set; (10) The direct superordinate element of the concept cmi.ckg.concept.parents: represents the direct superordinate set of the current query concept, access right: R, where the symbol "parents" represents the direct superordinate set; (11) The direct sub-set element of the concept cmi.ckg.concept.directHyponymSet: represents the direct sub-set of the current query concept, with access right: R, where the symbol "directHyponymSet" represents the direct sub-set; (12) Disturbance Set Element of Concept Relationship cmi.ckg.concept.relation.disturbSet: represents the disturbance set of concept relationship, where the disturbance is a relationship that is easily confused with the specified target relationship, access right: R, where: the symbol "disturbSet" represents the disturbance set; (13) Next learning concept element cmi.ckg.concept.nextConcept: represents the subsequent learning concept of the current query concept, access right: R, where: the symbol "nextConcept" represents the next concept; (14) The root node element cmi.ckg.root of the is-a classification structure: represents the root node of the is-a classification structure of the subject knowledge graph, with access right R, where: the symbol "root" represents the root node of the is-a classification structure; (15) All lower-level elements of the root node cmi.ckg.root.allHyponymNum: indicates the number of all lower-level nodes of the root node, access right: R, where: the symbol "allHyponymNum" indicates the number of all lower-level nodes; (16) Knowledge graph element cmi.ckg.KnowledgeGraph: represents the knowledge graph of the current query, access permission: R, where: the symbol "KnowledgeGraph" represents the knowledge graph; (17) The discipline element cmi.ckg.KnowledgeGraph.discipline of the knowledge graph: represents the discipline of the knowledge graph currently being queried, with access rights: R, where the symbol "discipline" represents the discipline; Sub-model 3.2 Communication data structure of learner cognitive state lcs: includes the following hierarchical communication elements: (1) Learner element cmi.lcs.learner: represents the current learner, access right: R, where the symbol "lcs" represents the learner cognitive state communication data structure, and "learner" represents the learner; (2) Learner identification code element cmi.lcs.learner.id: represents the identification code of the current learner, access right: R, where: the symbol "id" represents the identification code; (3) Learner's name element cmi.lcs.learner.name: indicates the name of the current learner, access rights: W+R, where: the symbol "name" indicates the name; (4) Current learning concept element cmi.lcs.concept: represents the learner's current learning concept, access right: R, where: the symbol "concept" represents the concept; (5) Concept learning assessment score element cmi.lcs.concept.AssessmentScore: represents the assessment score of the current learning concept, access rights: W+R, where: the symbol "AssessmentScore" represents the assessment score; (6) Assessment date of concept learning element cmi.lcs.concept.AssessmentDate: indicates the date of assessment of the current learning concept, access rights: W+R, where: the symbol "AssessmentDate" indicates the assessment date; (7) Concept cognitive prediction element cmi.lcs.concept.predict: represents cognitive prediction of the current learning concept, access rights: W+R, where: the symbol "predict" represents cognitive prediction; (8)Currently studied discipline element cmi.lcs.discipline: indicates the currently studied discipline, access right: R, where: the symbol "discipline" indicates the discipline; (9) The subject assessment score element cmi.lcs.discipline.AssessmentScore: represents the assessment score of subject learning, access rights: W+R, where: the symbol "AssessmentScore" represents the assessment score; (10) Discipline assessment date element cmi.lcs.discipline.AssessmentDate: indicates the date of discipline assessment, access rights: W+R, where: the symbol "AssessmentDate" indicates the assessment date; Sub-model 3.3 Distributed query communication interface of subject knowledge graph: expands the functions of the six API methods in the SCORM specification, specifically: (1) Initialize(): It is extended to activate the learning system LMS to initialize the communication dialogue with the KCO object, including creating a knowledge graph communication data structure cgg and a learner cognitive state communication data structure lcs for the communicating KCO object on the server side. Its interface format is: return_value = Initialize(parameter), where parameter parameter is an empty string, and return value return_value is a logical value "true" or "false", indicating whether the initialization communication dialogue is successful; (2) Terminate(): It is extended to terminate the communication dialogue of a KCO object, including releasing the knowledge graph communication data structure cgg and the learner cognitive state communication data structure lcs established for the KCO object. Its interface format is: return_value = Terminate(parameter), where parameter parameter is an empty string, and the return value return_value is a logical value "true" or "false", indicating whether the communication dialogue is terminated successfully; (3) SetValue(): It is extended to enable the KCO object to set the communication data elements in ckg and lcs. Its interface format is: return_value = SetValue (parameter1, parameter2), where parameter1 is the first string parameter, indicating the ckg and lcs communication data element to be set, parameter2 is the second string parameter, indicating the value of the communication data element specified by parameter1, and the return value return_value is a logical value "true" or "false", indicating whether the setting is successful; (4) GetValue(): extended to be able to pass the value of the communication data element specified in the ckg and lcs on the server side to the KCO object on the client side as the query result. Its interface format is: return_value = GetValue(parameter), where the parameter parameter represents the communication data element to be queried, and the return value return_value is the query result; (5) GetLastError(): extended to return the error code of the most recent access to ckg and lcs. Its interface format is: return_value = GetError(), where the return value symbol return_value receives the returned error code number; (6) GetErrorString(): extended to be able to return the error information corresponding to the error code of the most recent access to ckg and lcs. Its interface format is: return_value = GetErrorString(parameter), parameter parameter is the error code number, and the return value symbol return_value receives the returned error information string.

5. The interoperability model of the intelligent teaching system based on subject knowledge graph according to claim 4 is characterized by: The learner model described in sub-model 3 is based on the concept schema in the subject knowledge graph, and the concepts in the concept schema are used as the assessment points of the subject knowledge in the learner model.

6. The interoperability model of the intelligent teaching system based on subject knowledge graph according to claim 5 is characterized by: The sub-model 4 requires that KCO does not specify the subject knowledge graph to be operated in the design phase, and postpones the association between KCO and the subject knowledge graph it operates in the content packaging list until runtime. The specific implementation method is as follows; (1) Expand a knowledge graph element in the SCORM content packaging model <knowledge-graph> , used to describe the subject knowledge graph used in the content package;< / knowledge-graph> (2) Project elements in the SCORM content packaging model <item> In , an optional attribute knowledgeref of knowledge graph reference is extended to describe the association between the KCO object and the subject knowledge graph it references;< / item> (3) Resource elements in the SCORM content packaging model <resource> In, extend an optional attribute kcoimType describing the KCO interoperability model type and the corresponding attribute value "KCO";< / resource> (4) Extended <knowledge-graph> 、 <item>and <resource> The elements are defined in the XML schema file kcocp.xsd.< / resource> < / item> < / knowledge-graph> 7. The interoperability model of intelligent teaching system based on subject knowledge graph according to claim 6 is characterized by: The extended knowledge graph elements <knowledge-graph> The XML format is as follows:< / knowledge-graph> (1) Description: <knowledge-graph> Elements describe the subject knowledge graph used in content packaging;< / knowledge-graph> (2) Diversity: <knowledge-graph>Element in the manifest element <manifest> It may appear 0 or 1 times;< / manifest> (3) Attribute identifier: A unique identifier in the content manifest, which must be specified; (4) Attribute discipline: a string describing the discipline to which the discipline knowledge graph belongs, which must be specified; (5) Attribute href: a reference to the entry file of the subject knowledge graph, which must be specified; (6) Sub-elements <metadata> : Metadata describing the subject knowledge graph, optional element;< / metadata> (7) Sub-elements <file> : A list of files that the subject knowledge graph depends on, which must be specified.< / file>

Images