Online interactive learning system based on group intelligence

Abstract

The invention provides an online interactive learning system based on swarm intelligence. The system comprises a data memory, a search engine and a processor. The data memory is configured to store a course library comprising a plurality of course sets, a term library, a resource library, a user library participating in platform co-creation, and an interaction database in a user co-creation process. The system is characterized by comprising an automatic term extraction module, a man-machine interaction module, a term network construction module, a search module and a recommendation module, for the video resources in the course library, performing term extraction on the video resources based on an automatic term extraction module; a user supplements terms through an interactive interface, and term network construction is performed on a term set jointly generated by the user and a machine based on a term network construction module; and on the basis of the constructed term network, systematic search and recommendation services are established, and online interactive learning based on swarm intelligence i is realized. According to the technical scheme provided by the invention, at least three beneficial effects of deepening the coupling degree between the user and the learning platform, breaking the domain barrier and promoting the leading-edge domain communication and innovation can be realized, and knowledge generation, knowledge precipitation and knowledge creation of man-machine fusion are realized.

Description

Field of the Invention

[0001] The technical content disclosed by the present invention relates to the field of online education. Specifically, the technical solution disclosed by the present invention relates to a system for providing an immersive and interactive co-creation learning environment for users. Background of the Invention

[0002] With the development of Internet and multimedia technologies, online learning platforms have become a common choice. Generally, online learning platforms recommend relevant online course resources for users based on the course content learned by the users and user characteristics. Moreover, some current online learning platforms have leveraged artificial intelligence technologies to provide users with manual and semi-automatic note-taking and generation functions to assist users in learning and improve learning efficiency. However, current online learning platforms still have problems such as low coupling degree with users, fixed and homogeneous recommended content, and difficulty in capturing the forefront development of science. Summary of the Invention

[0003] Therefore, the present application proposes an online interactive learning system based on swarm intelligence to solve the problems such as low coupling degree with users and fixed and homogeneous recommended content existing in current online learning platforms. The online interactive learning system in the present application includes a data memory, a search engine, and a processor. The data memory is configured to store a course library containing multiple course sets, a term library, a resource library, a user library participating in platform co-creation, and an interaction database during user co-creation. The search engine is configured to search for resources related to the user query context in the memory. The processor is configured to present a corresponding interactive graphical interface via a user device.

[0004] Each course set in the course library, and each single lesson in each course set exhibits a unified format. Single courses with the same theme are placed in a course set, and it is confirmed by the knowledge of subject matter experts whether the courses belong to the same course set.

[0005] The resource library includes papers, books, website addresses, etc.

[0006] Users in the user library, as co-creators, can be divided into different levels according to the degree of interaction, and the degree of interaction can be evaluated according to certain rules based on the data in the interaction database. The evaluation method can be specified manually or based on the idea of attention flow (CN110245133A; Modeling collective attention in online and flexible learning environments; Distance Education 40:3, pages 303-308. (2019)).

[0007] This application proposes an online interactive learning system based on swarm intelligence, which includes five modules.

[0008] 1. Term Automatic Extraction Module: The term automatic extraction module refers to extracting terms from courses and resources through machine learning algorithms. It includes 1) Directly extract terms from the course; 2) First extract resources from the course, including scientific literature, books, and websites, etc., and then extract terms from the content of scientific literature, books, and websites.

[0009] 2. Human-Computer Interaction Module: The human-computer interaction module means that during the process of the user watching the video, bind the terms to the video playback time, and provide an interactive page for the user. The user can perform the following interactive operations according to their own knowledge 1) Modify, delete, or add the terms extracted by the term automatic extraction module; 2) Modify, delete, or add the resources extracted by the term automatic extraction module; 3) Add terms and resources that are not mentioned in the video resources but are related to the video resources.

[0010] 3. Term Network Construction Module: The term network construction module refers to using a certain term network construction algorithm to construct term networks for knowledge units at each level respectively. Knowledge units include single videos, video sets, resource sets corresponding to single videos, and resource sets corresponding to video sets. The specific process is as follows For a single video, perform the following operations 1) Obtain the term set Auto_DT directly extracted by the term automatic extraction module and the term set Auto_RT extracted from the resources; 2) Obtain the term set User_DT directly added by the user in the human-computer interaction module and the term set User_RT extracted from the resources added by the user; 3) Update the term set Auto_DT directly extracted by the term automatic extraction module according to the modification and deletion operations of the human-computer interaction module to obtain Update_Auto_DT; The term network of a single video is constructed based on Update_Auto_DT, Auto_RT, User_DT, and User_RT; The term network of the resources corresponding to a single video is constructed based on Auto_RT and User_RT; The term network of the video collection needs to merge Update_Auto_DT, Auto_RT, User_DT, and User_RT of each video in the video collection to obtain Multi_Update_Auto_DT, Multi_Auto_RT, Multi_User_DT, and Multi_User_RT, and is constructed based on Multi_Update_Auto_DT, Multi_Auto_RT, Multi_User_DT, and Multi_User_RT. The term network of the resource collection corresponding to the video collection is constructed based on Multi_Auto_RT and Multi_User_RT.

[0011] 4 Search module The search module refers to configuring a retrieval entry in the online interactive learning system based on swarm intelligence proposed in this application. After the user enters a retrieval term, the matching degree between the retrieval term and the term networks of knowledge units at different levels is analyzed, and the sorting results are displayed to the user in a certain form. Preferably, it can be displayed in units of course collections, and the individual courses with high relevance in the course collection are displayed. An icon is set for the displayed course collection and individual courses to show the relevance. And it can accurately recommend learning resources such as videos and resources at different levels to the user.

[0012] 5 Recommendation module The recommendation module mainly refers to calculating the relevance between the term network of the course on the current video player page and the term networks of other individual courses and course collections when the user views video resources. The ones with high relevance are displayed as recommendations around the video playback page.

[0013] Optionally, an online interactive learning system based on swarm intelligence proposed in this application may further include 1) A scientific literature automatic crawling module, which serves as a supplement to the system knowledge update. The user can also generate video content based on scientific literature; 2) A learning path automatic generation module. For a single video, the above-mentioned term automatic extraction module and human-computer interaction module extract its corresponding terms and resources, and automatically generate a learning path based on the terms and resources; 3) Provide the user with an interactive interface and functions such as outline positioning and subtitle display on the video playback page to facilitate learning. Beneficial effects

[0014] An online interactive learning system based on swarm intelligence proposed in this application can at least achieve three beneficial effects: deepening the coupling degree between users and the learning platform, breaking domain barriers, and promoting exchanges and innovations in frontier fields, so as to realize knowledge production, knowledge precipitation, and knowledge creation on the online interactive learning platform. Description of the Drawings

[0015] Figure 1 It is the logic diagram of each functional module of this application.

[0016] Figure 2 It is the flowchart of the embodiment of this application.

[0017] Figure 3 It is an implementation manner of the human-computer interaction interface described in this application.

[0018] Figure 4 It is an implementation manner of the video resource interface described in this application. Embodiment

[0019] In the online interactive learning system based on swarm intelligence proposed in this application, when there are at least 2 video sets, and each video set has at least 3 individual videos, the number of video sets is taken as 2, and the number of individual videos in each video set is 3. Denote the video sets and the corresponding individual videos as A{A1, A2, A3}, B{B4, B5, B6} respectively. As shown in the appendix Figure 1 The corresponding specific implementation steps are as follows: 1. For the individual videos in video sets A and B, extract the term sets. Based on OCR technology or speech recognition and natural language processing technology, extract the corresponding term sets DA1, DA2, DA3, DB4, DB5, DB6 and resource sets RA1, RA2, RA3, RB4, RB5, RB6 respectively, and obtain the corresponding term sets Auto_DTA1, Auto_DTA2, Auto_DTA3, Auto_DTB4, Auto_DTB5, Auto_DTB6; Take A1 as an example to illustrate the extraction process: 1.1 Use mature video processing tools such as FFmpeg and Opencv to extract frames from A1 respectively to obtain PA1; 1.2 Extract features of the images in PA1, including color histograms, texts, image edges, etc., and remove duplicate images to obtain De_PA1; 1.3 Use OCR technology to identify the content of the images in De_PA1, extract according to the characteristics of terms and documents, and obtain the term set Auto_DTA1 and the document set RA1; optionally, the text information of the video can also be extracted based on speech recognition technology, and then term extraction can be performed based on the text information.

[0020] 1.4 For the documents in the document set RA1, use the method in the patent "A Co-evolution Method of Text Classification and Term Network Growth" with the publication number CN114416997A to extract and obtain Auto_RTA1; 1.5 Merge Auto_DTA1 and Auto_RTA1 to obtain the initial term set Auto_TA1 of A1 and construct the term network Auto_TNA1 of video A1. Construct the term networks Auto_TNA2, Auto_TNA3, Auto_TNB4, Auto_TNB5, Auto_TNB6 of each video in video set A and video set B in accordance with this method. Merge Auto_DTA1, Auto_DTA2, Auto_DTA3 to obtain the term set Auto_DTA of video set A and construct the term network Auto_DTNA of video set A; merge Auto_DTB4, Auto_DTB5, Auto_DTB6 to obtain the term set Auto_TB of video set B and construct the term network Auto_DTNB of video set B; corresponding Figure 1 to the operations of initialization, the first warehousing, and constructing the term network; 1.6 Based on the term network in step 1.5, the similarity of the term networks of individual videos and video sets can be preliminarily calculated, and the resource recommendation content of individual video pages and video set pages can be configured. There are many studies in the field of graph similarity calculation currently, including methods such as Graph embedding, Pairwise Node Comparison, Graph Matching Networks, etc., which can be selected according to requirements in specific scenarios.

[0021] 2 Human-computer interaction, update the term library and term network. On the video playback page, match the terms in the extracted term set with the timeline. When the video plays to the corresponding time, display the terms and resources automatically recognized for the current content. Users can modify, delete, or add terms and resources. Based on the user's deletion and modification actions, update the term sets Update_Auto_DTA1, Update_Auto_DTA2, Update_Auto_DTA3, Update_Auto_DTB4, Update_Auto_DTB5, Update_Auto_DTB6 for individual videos and video sets; based on the user's addition actions, obtain the term sets User_DTA1, User_DTA2, User_DTA3, User_DTB4, User_DTB5, User_DTB6 co-created by the user and the system, and the term sets User_RTA1, User_RTA2, User_RTA3, User_RTB1, User_RTB2, User_RTB3 corresponding to the resources. Construct the updated term network Update_Auto_TNA2, Update_Auto_TNA3, Update_Auto_TNB4, Update_Auto_TNB5, Update_Auto_TNB6 for each video; merge Update_Auto_DTA1, Update_Auto_DTA2, Update_Auto_DTA3 to obtain the term set Update_Auto_DTA for video set A, and construct the term network Update_Auto_DTNA for video set A; merge Update_Auto_DTB4, Update_Auto_DTB5, Update_Auto_DTB6 to obtain the term set Update_Auto_DTB for video set B, and construct the term network Update_Auto_DTNB for video set B; corresponding to Figure 1 the operations of warehousing for the second and subsequent times and constructing the term network.

[0022] 3 Based on the term network in step 2, update the similarity of the term networks for individual videos and video sets, and configure the resource recommendation content for the individual video page and video set page.

[0023] 4 On the retrieval interface, still match the above-constructed term network with the retrieval content input by the user, and recommend the corresponding individual video or video set content to the user.

[0024] It should be noted that an online interactive learning system based on swarm intelligence protected by this application is a solution that can integrate humans and machines to achieve knowledge generation, knowledge precipitation, and knowledge creation. The video processing technology, speech processing technology, and term extraction technology therein can be updated according to the evolution of the technology.

Claims

1. An online interactive learning system based on swarm intelligence, the system comprising a data storage, a search engine and a processor. The data storage is configured to store a course library containing a plurality of course sets, a term library, a resource library, a user library participating in platform co-creation and an interactive database in the process of user co-creation. It is characterized by The system includes a term automatic extraction module, a human-computer interaction module, a term network construction module, a search module, and a recommendation module. For video resources in the course library, the term automatic extraction module extracts terminology from the video resources. Users supplement terminology through the interactive interface, and the term network construction module constructs a term network for the term set jointly generated by users and machines. Based on the constructed terminology network, a systematic search and recommendation service is built to achieve online interactive learning based on swarm intelligence.

2. The online interactive learning system as claimed in claim 1, characterized in that The automatic term extraction module is to extract terms from courses and resources through machine learning algorithms, including 1) Extracting terms directly from the course; 2) First extract resources from the course, including scientific literature, books, and websites, and then extract terms from the content of scientific literature, books, and websites.

3. The online interactive learning system as claimed in claim 1, characterized in that The human-computer interaction module refers to binding the terms with the video playing time during the user's video viewing process, and providing the user with an interactive page. The user can perform the following interactive operations based on his or her own knowledge. 1) Modify, delete or add terms extracted in the automatic term extraction module; 2) Modify, delete or add the resources extracted in the terminology automatic extraction module; 3) Add terms and resources that are not mentioned in the video resources but are related to the video resources.

4. The online interactive learning system as claimed in claim 1, characterized in that The term network construction module refers to the use of a certain term network construction algorithm to construct a term network for each level of knowledge units. The knowledge units include a single video, a video collection, a resource collection corresponding to a single video, and a resource collection corresponding to a video collection. The specific process is as follows: For a single video, perform the following operations 1) Obtain the term set Auto_DT directly extracted by the automatic term extraction module and the term set Auto_RT extracted from the resource; 2) Obtain the term set User_DT directly added by the user of the human-computer interaction module and the term set User_RT extracted from the user-added resources; 3) According to the modification and deletion operations of the human-computer interaction module, the term set Auto_DT directly extracted by the term automatic extraction module is updated to obtain Update_Auto_DT; The term network of a single video is built based on Update_Auto_DT, Auto_RT, User_DT, and User_RT; the resource term network corresponding to a single video is built based on Auto_RT and User_RT; the term network of a video collection needs to merge the Update_Auto_DT, Auto_RT, User_DT, and User_RT of each video in the video collection to obtain Multi_Update_Auto_DT, Multi_Auto_RT, Multi_User_DT, and Multi_User_RT, and build it based on Multi_Update_Auto_DT, Multi_Auto_RT, Multi_User_DT, and Multi_User_RT; the term network of the resource collection corresponding to the video collection is built based on Multi_Auto_RT and Multi_User_RT.

5. The online interactive learning system as claimed in claim 1, characterized in that The search module refers to a search entry configured in the online interactive learning system based on swarm intelligence proposed in this application. After the user enters the search term, the matching degree between the search term and the term network of knowledge units at different levels is analyzed, and the sorting results are displayed to the user in a certain form. Preferably, the display can be performed in units of course sets, and the single course with high relevance in the course set is displayed. An icon is set for the displayed course set and single course to show the relevance. It can accurately recommend learning resources such as videos and resources at different levels to users.

6. The online interactive learning system as claimed in claim 1, characterized in that The recommendation module mainly refers to calculating the relevance of the term network of the course on the current video player page with the term network of other single courses and course sets when the user is watching a video resource, and displaying recommendations with high relevance around the video playback page.

Images