A content reservation and recommendation method for AI online education

Abstract

The present invention discloses a content reservation and recommendation method for AI online education, comprising: for recorded courses or delayed live courses, a server sends the original video stream to all student terminals in sequence through a preset number of other routers and a gateway router, and at the same time sends generated content directory information and preset recommendation information to all student terminals, wherein the gateway router has the functions of a gateway and a router; the method reserves content points based on the recommendation information and adaptively adjusts the playback speed during the course playback, thereby improving the personalization and efficiency of learning; after each reserved content is played, the method provides questions for students to answer, and by judging the accuracy of the answers and sending detailed video streams to the student terminals, the student's learning effect is improved and interactive feedback is achieved.

Description

Technical Field

[0001] The present invention belongs to the field of online education technology, and specifically relates to a content reservation and recommendation method for AI online education. Background Art

[0002] In online education scenarios, each student focuses on different course content and hopes to achieve personalized education. That is, it is necessary to target the content concerns of different students to achieve course content positioning, speed adjustment, understanding level assessment, and detailed explanation of key content as needed.

[0003] Recorded or live courses also provide a course summary and table of contents to help students better understand the course. Students can schedule entire courses, but they cannot schedule specific content within a course. There is no support for intelligent speed adjustment during playback based on student interest. There is no ability to schedule course content at a more granular level, tailored to individual student needs, or for adaptive playback based on schedules, resulting in a lack of flexibility and personalization in course viewing. Furthermore, there is no support for interactive feedback on content points or on-demand detailed explanations of key points. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems raised in the background technology and propose an AI online education content reservation and recommendation method.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] The present invention proposes an AI online education content reservation and recommendation method, comprising:

[0007] For recorded or delayed live courses, the server sends the original video stream to all student terminals through a preset number of other routers and gateway routers in sequence, while also sending the generated content directory information and preset recommendation information to all student terminals. The gateway router has the functions of both a gateway and a router.

[0008] Each student can reserve content based on preset recommendation information on the corresponding student terminal as needed;

[0009] When playing a course, each student terminal detects whether it is a reserved content and adjusts the playback speed adaptively;

[0010] After each student terminal finishes playing a content point, the server sends questions related to the current content point to the student terminal for the student to answer. The student terminal pauses the playback of the original video stream and determines whether the student's answer accuracy reaches a first threshold. If the first threshold is reached, the original video stream continues to play. If the first threshold is not reached, a request for a detailed explanation video stream of the corresponding content point is sent to the server, and the gateway router stops sending the original video stream to the student terminal that has not reached the first threshold.

[0011] After receiving the detailed explanation video stream request for each content point, the server sends a first response message to the corresponding student terminal;

[0012] Each corresponding student terminal parses the received first response information, and when the status of the first response information is acceptance, sends a corresponding detailed video stream request message to the server according to the parsed content;

[0013] The server then forwards the detailed video stream to the gateway router of the corresponding student terminal through a preset number of other routers;

[0014] The gateway router of each student terminal sends the detailed video stream to the student terminal that has not reached the first threshold, and caches the original video stream. When the cache space usage of the gateway router is greater than the third threshold, it sends a cache request message to other routers upstream of it.

[0015] Preferably, the content directory information includes recorded content directory information and live content directory information;

[0016] For recorded courses, the complete audio of the course is identified and content points and key content points are marked to form recorded content directory information. The recorded content directory information includes: directory ID, course ID, course type, creation time, content point sequence number, content point name, content point type, content point description, content point start timestamp, content point end timestamp, and whether each content point is a key content point;

[0017] For delayed live courses, the teacher's course introduction is identified and content points and key content points are marked to form live content directory information. The live content directory information includes: directory ID, course ID, course type, creation time, content point sequence number, content point name, content point type, content point description, and whether each content point is a key content point.

[0018] Preferably, each student makes a content reservation based on preset recommendation information on the corresponding student terminal as needed, including:

[0019] For recorded courses, the preset recommendation information includes at least three recommendation information: recommendation based on the student's own historical reservation content point preference, recommendation based on key content points, and recommendation based on the historical playback content points of other students.

[0020] For delayed live courses, the preset recommendation information includes at least three recommendation information: recommendation based on the student's own historical reservation content point preference, recommendation based on key content points, and recommendation based on real-time reservation information of other students;

[0021] Each student can make or not make a reservation for content according to the recommended information as needed;

[0022] Among them, for recommendations based on real-time reservation information from other student terminals, each student terminal sends the reservation content point information to the teacher terminal and all other student terminals, and the student terminal counts the received reservation content point information, calculates the reservation rate of each content point, and recommends the content point whose reservation rate exceeds the preset recommendation threshold as real-time reservation information for other student terminals.

[0023] Preferably, when playing a recorded course or a delayed live course, each student terminal detects whether it is a reserved content and performs adaptive adjustment of the playing speed, including:

[0024] For recorded courses, the student terminal detects the current playback time point. If the current playback time point is between the start and end timestamps of the scheduled content, the playback is slowed down according to the first preset speed. Otherwise, the playback is accelerated according to the second preset speed.

[0025] For delayed live courses, the student client dynamically extracts the audio of the currently playing video stream and performs text extraction. Using natural language processing technology, the extracted text and the current student reservation content are vectorized and the difference between the two vectors is calculated. If the difference is less than a second threshold, the video is played at a reduced speed according to the first preset speed. Otherwise, the video is played at a accelerated speed according to the second preset speed.

[0026] If a student has not reserved any content points, the corresponding student terminal will adaptively adjust the playback speed according to the recommendations for key content points;

[0027] For recorded courses, when students have not made reservations for all content points and there are no key content points, the corresponding student terminal will adaptively adjust the playback speed based on the recommendations of historical playback content points on other student terminals.

[0028] Preferably, the detailed explanation video stream request includes a message ID, a message type, a student ID, a creation time, a course ID, and a content point detailed explanation video stream ID;

[0029] The first response information includes message ID, message type, student ID, creation time, response status, multicast group address, original video stream port and detailed video stream port;

[0030] The fields of the detailed video stream request message include message type, checksum, multicast group address, original video stream port and detailed video stream port, wherein the student end adds the original video stream port and detailed video stream port in the parsed first response information to the IGMP v2 message to form a detailed video stream request message.

[0031] Preferably, when the student terminal sends a detailed video stream request message to the server, it is forwarded to the server through the gateway server and other routers. After the gateway server receives the detailed video stream request message, it parses it and records the original video stream port and the detailed video stream port. The gateway router receives the original video stream and the detailed video stream, and distinguishes the original video stream and the detailed video stream through the target port number information of the video stream according to the ACL access control list rules on the gateway router.

[0032] Preferably, each student terminal and the server include a gateway router located at the student terminal and n other routers, and the n other routers are sequentially represented as a first router P1, a second router P2, ..., an nth router Pn from the gateway router to the server;

[0033] When caching the original video stream, the gateway router records the time point t1 at which the original video stream is cached and stores it in the mapping table, and continuously monitors whether the cache space usage rate is greater than a third threshold. When the cache space usage rate is greater than the third threshold, the gateway router sends a cache request message to the first router P1 upstream of the gateway router.

[0034] After receiving the cache request message, the first router P1 decides whether to accept the cache request based on its own cache space;

[0035] When the first router P1 accepts the cache request, it sends a second response message of acceptance to the gateway router, and the gateway router records the IP address of the first router P1 in the second response message, and records the time point t2 when the first router P1 starts caching the original video stream and stores it in the mapping table. The gateway router itself stops caching the original video stream, and the first router P1 starts caching the original video stream at time point t2.

[0036] When the first router P1 rejects the cache request, it sends a second response message of rejection to the gateway router, and the gateway router stops sending the detailed video stream to the student terminal that does not reach the first threshold, and resends the original video stream to the student terminal that does not reach the first threshold.

[0037] Preferably, when caching the original video stream, the first router P1 continuously monitors whether the cache space usage is greater than a third threshold, and sends a cache request message to the second router P2 adjacent upstream thereof when the usage is greater than the third threshold;

[0038] After receiving the cache request message, the second router P2 decides whether to accept the cache request based on its own cache space;

[0039] When the second router P2 accepts the cache request, it simultaneously sends a third response message of acceptance to the first router P1 and the gateway router adjacent to it downstream. The gateway router records the IP address of the second router P2 in the third response message, as well as the time point t3 when the second router P2 starts caching the original video stream, and stores the data in the mapping table. The first router P1 stops caching the original video stream, and the second router P2 starts caching the original video stream at time t3.

[0040] When the second router P2 rejects the cache request, it only sends a third response message of rejection to its adjacent downstream first router P1, and the first router P1 sends a second response message of rejection to the gateway router. At this time, the gateway router stops sending the detailed video stream to the student terminal that does not reach the first threshold, and resends the original video stream to the student terminal that does not reach the first threshold;

[0041] When the cache space usage rate of one of the n other routers Px is always less than the third threshold, the gateway router stops sending cache request messages to its other upstream routers, and records the time point t(x+1) when the other router Px starts caching the original video stream and stores it in the mapping table;

[0042] The mapping table stores a mapping relationship between each time point at which the original video stream starts to be cached and the cache device, and the cache device is each router.

[0043] Preferably, the content reservation and recommendation method for AI online education further comprises: when the detailed video stream request is finished playing on the student terminal, the corresponding student terminal sends the original video stream request information to the gateway router;

[0044] After receiving the original video stream request information, the gateway router selects the earliest time point t1 to record the original video stream according to the mapping table, and sends the original video stream starting at time t1 to the corresponding student terminal;

[0045] When the gateway router sends the original video stream to time point t2, according to the mapping table, the gateway router sends the original video stream request information to the cache device first router P1 corresponding to the cache start time point t2;

[0046] After receiving the original video stream request information, the first router P1 forwards the original video stream starting at time t2 to the corresponding student terminal through the gateway router;

[0047] When the network router forwards the original video stream to time point t3, according to the mapping table, the network router sends the original video stream request information to the cache device second router P2 corresponding to the cache start time point t3;

[0048] After receiving the original video stream request information, the second router P2 forwards the original video stream starting at time t3 to the corresponding student terminal through the gateway router;

[0049] And so on, until other routers Px receive the original video stream request information, they forward it to the corresponding student terminal through the gateway router and send the original video stream with the starting time point t(x+1).

[0050] Preferably, the content reservation and recommendation method for AI online education further comprises: after each of the routers sends the original video stream to the corresponding student terminal, clearing the cache of the sent original video stream;

[0051] After the gateway router clears the cache, when it detects that the cache space usage of the gateway router is less than a fourth threshold, it sends a stop caching request message to other routers Px that are caching the latest data of the original video stream;

[0052] After receiving the stop caching request message from the gateway router, if the other router Px accepts the stop caching, the other router Px stops caching the original video stream and sends a stop caching request response message with a response status of accepted to the gateway router;

[0053] After receiving the accepted stop caching request response message, the gateway router restarts caching the remaining uncached original video streams.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] 1. This method improves the personalization and efficiency of learning by reserving content points based on recommendation information and adaptively adjusting the playback speed during the course playback;

[0056] 2. After each scheduled content is played, this method provides questions for students to answer, judges the accuracy of the answers, and sends a detailed video stream to the student end to improve students' learning effects and achieve interactive feedback;

[0057] 3. For students who need detailed explanations, this method allows each router to cache the original video data while forwarding the detailed explanation video. This ensures that students can seamlessly continue watching the original video after watching the detailed explanation video, improving the continuity and fluency of learning. In addition, each router can send cache requests to upstream routers based on cache space usage, realizing dynamic allocation and optimization of cache resources.

[0058] 4. This method is based on IGMPv2 messages, and adds new fields to obtain detailed video stream request messages. The new fields are used to identify the specific content points of the request, thereby enhancing the flexibility and practicality of the protocol; and by identifying the specific content points of the request, the flexibility and practicality of the protocol are enhanced. The corresponding video streams are sent separately, which not only retains the efficiency of multicast and reduces network complexity, but also enables the gateway router to distinguish different video streams, which can avoid the limitation of all members receiving the same content in traditional multicast and provide technical support for personalized learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Flowchart of the content reservation and recommendation method for AI online education of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0061] like Figure 1 As shown in the figure, a content reservation and recommendation method for AI online education is provided. First of all, it should be explained that recorded courses refer to pre-recorded video courses that students can watch or listen to at any time and any place through the Internet; delayed live courses refer to live courses where the server sends the original video stream with a slight delay (e.g., 5 minutes). Students can watch the course in near real time and can cache unplayed content locally for accelerated playback.

[0062] The content reservation and recommendation methods of this AI online education include:

[0063] Step 1: For recorded or delayed live courses, the server sends the original video stream to all student terminals through a preset number of other routers and gateway routers in sequence (i.e., the server sends the original video stream to all student terminals through routers at all levels of the network link between the server and the student terminals (including the gateway router on the student terminal side and other upstream routers)), and sends the generated content directory information and preset recommendation information to all student terminals. The gateway router has the functions of both a gateway and a router.

[0064] Among them, the recorded courses or delayed live courses on the teacher's side send video streams through the server in multicast mode, and each student side receives the video stream for playback; between each student side and the server, there is a gateway router located at the student side, and n other routers, and the n other routers are represented in sequence from the gateway router to the server as the first router P1, the second router P2,..., the nth router Pn (the other routers adjacent upstream of a certain other router are other routers adjacent to the certain other router and close to the server side).

[0065] The content directory information includes recorded content directory information (statically generated in advance) and live content directory information (dynamically generated in real time);

[0066] For recorded courses, the complete audio of the course is recognized in advance (text can be extracted through voice recognition technology and content points and key content points can be extracted through natural language processing technology, which is a prior art and will not be described in detail) and content points and key content points are marked (wherein, the identification of key content points refers to the teacher's special emphasis on key content during the explanation. For example, if the teacher says "Next, we will introduce the working principle of the large model. This part is very critical. Please pay attention to the lecture", then "The working principle of the large model" will not only be identified as a content point, but also marked as a key content point). The recorded content directory information is formed, and the recorded content directory information includes: directory ID, course ID, course type (delayed live course or recorded course), creation time, content point sequence number, content point name, content point type, content point description, content point start timestamp, content point end timestamp, and whether each content point is a key content point;

[0067] For delayed live courses, the teacher's course introduction is identified and content points and key content points are marked (the teacher first gives a general introduction to the course content after the delayed live course begins (at this time, you can first click the "Voice Recording" button in the course introduction on the teacher side, and then click "End Recording" after the general introduction to complete the segment marking of the introduction part)), forming the live content directory information (for example, at the beginning of the live course, the teacher introduces "Today's course has three parts. First, we will introduce the basic concepts of the big model, then the working principles of the big model, and finally the application scenarios of the big model. Among them, the working principles of the big model are the key content", then the content directory information is marked as: 1. Basic concepts of the big model; 2. Working principles of the big model (key content); 3. Application scenarios of the big model). The live content directory information also includes: directory ID, course ID, course type, creation time, content point sequence number, content point name, content point type (such as basic concepts, technical principles, development trends, application scenarios, etc.), content point description, and whether each content point is a key content point.

[0068] Step 2: Each student makes a content reservation based on preset recommendation information on the corresponding student terminal as needed, including:

[0069] For recorded courses, the preset recommendation information includes at least three types of recommendation information: recommendation based on the student's own historical reservation content point preferences, recommendation based on key content points, and recommendation based on the content points of other students' historical playback;

[0070] For delayed live courses, the preset recommendation information includes at least three types of recommendation information: recommendations based on the student's own historical reservation content point preferences, recommendations based on key content points, and recommendations based on real-time reservation information of other students.

[0071] Each student can make a content reservation (multiple or one) or not make a reservation according to the recommended information (on their respective student terminals) as needed;

[0072] Recommendations based on students' past content preferences are analyzed based on their preferences. Data profiling is used to label students' learning preferences, such as basic concepts, technical principles, development trends, and application scenarios. The content type of the course content point is matched to the student's learning preference tags, and matching content points are recommended to students.

[0073] Among them, recommendations are based on key content points: that is, the content points that teachers emphasize.

[0074] Among them, recommendations are based on historical playback content points from other student terminals: among all content points in the current recorded course, content points whose historical playback rate (the ratio of the number of students who have played back the content point to the total number of students who have historically watched the recorded course) exceeds the recommendation threshold (such as 50%) are recommended.

[0075] Among them, for recommendations based on real-time reservation information from other student terminals, each student terminal will send the reservation content point information to the teacher terminal and all other student terminals (and receive content reservation information from other student terminals at the same time), and the student terminal will count the received reservation content point information, dynamically calculate the reservation rate of each content point (that is, the ratio of the number of students who have reserved the current content point to the total number of students), and recommend content points with a reservation rate exceeding the preset recommendation threshold (such as 40%) as real-time reservation information for other student terminals; among them, after receiving and summarizing the content reservation information from all student terminals, the teacher terminal will display the reservation status (that is, display the current reservation rate of each content point). The teacher can understand the overall attention of students to the content points in this course, and can adjust the level of detail of the content explanation in subsequent explanations, for example, giving a more detailed introduction to content points with high reservation rates, and appropriately simplifying the content points with low reservation rates and non-important knowledge points.

[0076] It should be noted that after this reservation, students can also make or cancel reservations for content points that have not yet reached the playback point during the subsequent course viewing process (content points that have already been played cannot be canceled).

[0077] Step 3: When playing a course (recorded or delayed live), each student client detects whether it is a reserved content and adaptively adjusts the playback speed (when the student client initially watches the course, it plays at normal speed, i.e. 1x speed), including:

[0078] For recorded courses, the student terminal detects the current playback time point. If the current playback time point is between the start and end timestamps of the scheduled content, the playback is slowed down according to the first preset speed (such as 0.9 times the speed). Otherwise, the playback is accelerated according to the second preset speed (such as 1.1 times the speed).

[0079] For delayed live courses, the student client dynamically extracts the audio of the currently playing video stream and performs text extraction (text extraction can be performed using voice recognition technology). Using natural language processing technology, the extracted text and the current student client's appointment content (for each student client, the appointment content on each student client is the appointment made by the corresponding student) are vectorized and the difference between the two vectors is calculated. If the difference is less than a second threshold (the smaller the difference, the higher the similarity between the two vectors), the video is played at a reduced speed according to the first preset speed. Otherwise, the video is played at a accelerated speed according to the second preset speed.

[0080] If the student has not reserved any content points, the corresponding student terminal will perform adaptive adjustment of the playback speed according to the recommendation of key content points (the specific adaptive adjustment process is the same as the adaptive adjustment in step 3);

[0081] For recorded courses, when students have not made reservations for all content points and there are no key content points, the corresponding student terminal will adaptively adjust the playback speed based on the recommendations of historical playback content points on other student terminals (the specific adaptive adjustment process is the same as the adaptive adjustment in step 3).

[0082] Step 4: For recorded or delayed live courses, after each student finishes playing a content point, the server sends questions related to the current content point to the student for them to answer (the number of questions can be set as needed, such as 3-5, and the questions will appear in a pop-up window on the student). The student pauses playback of the original video stream and determines whether the student's answer accuracy reaches a first threshold (e.g., 60%). If so, playback of the original video stream continues. If not, a request for a detailed explanation video stream for the corresponding content point is sent to the server. If the student determines that the answer accuracy is below the first threshold, the student will be prompted with the message "Since your answer accuracy for the current content point is less than 60%, the system will provide a detailed explanation of this part!" After the student confirms the answer, the student sends a detailed explanation video stream request for the corresponding content point to the server. The gateway router stops sending the original video stream to students that have not reached the first threshold (and caches the original video stream). The detailed explanation video stream request includes the message ID, message type, student ID, creation time, course ID, and content point detailed explanation video stream ID.

[0083] It should be noted that the transmission of the detailed video stream request and the subsequent first response information detailed video stream request message between the server and the student end is forwarded through the gateway router and other routers.

[0084] Step 5: After receiving the detailed explanation video stream request for each content point, the server sends a first response message to the corresponding student terminal; the first response message includes the message ID, message type, student ID, creation time, response status (accepted or rejected), multicast group address, original video stream port, and detailed explanation video stream port;

[0085] Among them, when the response status in the first response information is accepted, an accepted response information is sent (the response status is 1) to notify the student end that it will send the original video stream and the detailed explanation video stream using the multicast group address G. The original video stream uses port Port1 (such as 5004), and the detailed explanation video stream uses port Port2 (such as 5005); afterwards, the server uses the same multicast group address (G) but different UDP ports to send two streams according to the first response information. The original normal playback video uses Port1, and the detailed explanation video uses Port2.

[0086] When the response status in the first response information is rejection, a rejection response information (response status is 0) is sent to notify the student end to reject the detailed video stream request. In the first response information, the multicast group address, original video stream port, and detailed video stream port are all set to 0.

[0087] Step 6: Each corresponding student terminal parses the received first response information, and when the status of the first response information is accepted, sends a corresponding detailed video stream request message to the server according to the parsed content;

[0088] Among them, the fields of the detailed video stream request message include message type (i.e. Type, which is an existing field in the IGMPv2 message. A new type is added to the IGMP v2 message, and the value is 0x18), checksum (i.e. Checksum, which is an existing field in the IGMPv2 message. Checksum, the message checksum is calculated and filled in the field), multicast group address (i.e. Group Address, which is an existing field in the IGMPv2 message), original video stream port (i.e. Original VideoPort, a newly added field) and detailed video stream port (i.e. Detailed VideoPort, a newly added field). The student end adds the original video stream port and detailed video stream port in the first response information after parsing to the IGMP v2 message to form a detailed video stream request message (IGMPv2: IGMPv2 (Internet Group Management Protocol version 2) A network protocol used by hosts (referring to students) and adjacent routers to communicate with each other to report which multicast groups they wish to join or leave, allowing routers to determine which multicast traffic to forward on the network. It manages multicast group membership by sending IGMP query and report messages, ensuring that only interested receivers receive multicast data.

[0089] Among them, when the student end sends a detailed video stream request message to the server, it is forwarded to the server through the gateway server and other routers. After the gateway server receives the detailed video stream request message, it parses it and records the original video stream port and the detailed video stream port (that is: after the gateway router receives the detailed video stream request message from the student end, it parses it and updates its multicast forwarding list, recording the multicast group address as G, the original video stream port as Port1, and the detailed video stream port as Port2).

[0090] Step 7: After receiving the detailed video stream request message, the server forwards the detailed video stream to the gateway router of the corresponding student terminal through a preset number of other routers;

[0091] It should be noted that the gateway router and other routers use the multicast routing protocol to maintain the forwarding routing information of the multicast group. The detailed video stream with the multicast group address G sent by the server is forwarded by other routers through the routing information and finally sent to the gateway router (the original video stream is also forwarded by other routers through the routing information and finally sent to the gateway router).

[0092] The gateway router receives the original video stream and the detailed video stream, and distinguishes the original video stream and the detailed video stream according to the video stream port in the ACL access control list rule (after the gateway router receives the original video stream and the detailed video stream, it identifies and distinguishes the original video stream and the detailed video stream through the ACL rule (server IP and video stream port) based on the five-tuple (source IP address, destination IP address, source port number, destination port number, protocol type) filtering method).

[0093] Step 8. The gateway router of each student end sends a detailed video stream to the student end that has not reached the first threshold (to the student end corresponding to the student that has not reached the first threshold) and caches the original video stream. When the cache space usage rate of the gateway router is greater than the third threshold, it sends a cache request message to other routers upstream of it.

[0094] When caching the original video stream, the gateway router records the time point t1 at which the original video stream starts to be cached and stores it in the mapping table. The gateway router also continuously monitors whether the cache space usage rate is greater than a third threshold (e.g., 70%). If the cache space usage rate is greater than the third threshold, the gateway router sends a cache request message to the first router P1 upstream of the gateway router.

[0095] After receiving the cache request message, the first router P1 decides whether to accept the cache request based on its own cache space (e.g., if its own cache space is already greater than the third threshold, the cache request will be rejected);

[0096] When the first router P1 accepts the cache request, it sends a second response message of acceptance to the gateway router, and the gateway router records the IP address of the first router P1 in the second response message, and records the time point t2 when the first router P1 starts caching the original video stream and stores it in the mapping table. The gateway router itself stops caching the original video stream, and the first router P1 starts caching the original video stream at time point t2.

[0097] When the first router P1 rejects the cache request, it sends a second response message of rejection to the gateway router, and the gateway router stops sending the detailed explanation video stream to the student terminal that has not reached the first threshold, and resends the original video stream to the student terminal that has not reached the first threshold (and sends the reason, such as "Because the video stream cannot be cached anymore, in order to ensure the normal playback of the course, the detailed explanation video stream is interrupted this time");

[0098] Similarly, when caching the original video stream, the first router P1 continuously monitors whether the cache space usage is greater than the third threshold, and sends a cache request message to the second router P2 adjacent upstream when the usage is greater than the third threshold;

[0099] After receiving the cache request message, the second router P2 decides whether to accept the cache request based on its own cache space;

[0100] When the second router P2 accepts the cache request, it simultaneously sends a third response message of acceptance to the first router P1 and the gateway router adjacent to it downstream. The gateway router records the IP address of the second router P2 in the third response message, as well as the time point t3 when the second router P2 starts caching the original video stream, and stores the data in the mapping table. The first router P1 stops caching the original video stream, and the second router P2 starts caching the original video stream at time t3.

[0101] When the second router P2 rejects the cache request, it only sends a third response message of rejection to its adjacent downstream first router P1, and the first router P1 sends a second response message of rejection to the gateway router. At this time, the gateway router stops sending the detailed video stream to the student terminal that does not reach the first threshold, and resends the original video stream to the student terminal that does not reach the first threshold;

[0102] This process continues in this way until the cache space usage rate of one of the n other routers, Px, is consistently lower than the third threshold. The gateway router then stops sending cache request messages to its other upstream routers, and records the time t(x+1) when the other router Px starts caching the original video stream and stores it in the mapping table. (In this embodiment, when the second router P2 is caching the original video stream, the cache space usage rate is not higher than the third threshold, meaning that the original video stream has been cached in the second router P2.)

[0103] The mapping table stores the mapping relationship between each time point when the original video stream starts to be cached and the cache device, and the cache device is each router.

[0104] The content reservation and recommendation method for AI online education further includes: when the detailed explanation video stream request is finished playing on the student terminal (or the student terminal actively exits the detailed explanation video playback of the content point), the corresponding student terminal sends the original video stream request information to the gateway router;

[0105] After receiving the original video stream request information, the gateway router selects the earliest time point t1 to record the original video stream according to the mapping table, and sends the original video stream starting at time t1 to the corresponding student terminal;

[0106] When the gateway router sends the original video stream to time point t2, according to the mapping table, the gateway router sends the original video stream request information to the cache device first router P1 corresponding to the cache start time point t2;

[0107] After receiving the original video stream request information, the first router P1 forwards the original video stream starting at time t2 to the corresponding student terminal through the gateway router;

[0108] When the gateway router forwards the original video stream to time point t3, according to the mapping table, the gateway router sends the original video stream request information to the cache device second router P2 corresponding to the cache start time point t3;

[0109] After receiving the original video stream request information, the second router P2 forwards the original video stream starting at time t3 to the corresponding student terminal through the gateway router;

[0110] And so on, until other routers Px receive the original video stream request information, they forward it to the corresponding student terminal through the gateway router and send the original video stream with the starting time point t(x+1).

[0111] The content reservation and recommendation method for AI online education further includes: after each router (including a gateway router and other routers) sends an original video stream to a corresponding student terminal, clearing a cache of the sent original video stream;

[0112] After the gateway router clears the cache, when it detects that the cache space usage of the gateway router is less than the fourth threshold (where the fourth threshold is less than the third threshold), it sends a stop caching request message to the other router Px that is caching the latest data of the original video stream (in this embodiment, the stop caching request message is sent to the second router P2);

[0113] After receiving the stop caching request message from the gateway router, if the other router Px accepts the stop caching, the other router Px stops caching the original video stream and sends a stop caching request response message with a response status of accepted to the gateway router;

[0114] After receiving the accepted stop caching request response message, the gateway router restarts caching the remaining uncached original video streams.

[0115] Finally, return to step 3 and continue to follow the steps in sequence.

[0116] In another embodiment, the present application also includes an AI online education content reservation and recommendation device, including a processor and a memory storing a plurality of computer instructions, which implement steps 1 to 3 of the method when the computer instructions are executed by the processor.

[0117] This method makes reservations for content points based on recommendation information and adaptively adjusts the playback speed during the course playback, thereby improving the personalization and efficiency of learning; after each reserved content is played, this method provides questions for students to answer, and judges the accuracy of the answers and sends detailed explanation video streams to the student end to improve students' learning effects and achieve interactive feedback; for students who need detailed explanations, this method caches the original video data while forwarding the detailed explanation video, ensuring that students can seamlessly continue to watch the original video after watching the detailed explanation video, thereby improving the continuity and fluency of learning. In addition, each router The router can also send cache requests to upstream routers based on the usage of cache space, thereby realizing dynamic allocation and optimization of cache resources; this method is based on IGMPv2 messages, and a new field is added to obtain a detailed video stream request message, and the specific content points of the request are identified by the new field, thereby enhancing the flexibility and practicality of the protocol; and by identifying the specific content points of the request, the flexibility and practicality of the protocol are enhanced. The corresponding video streams are sent separately, which not only retains the efficiency of multicast and reduces network complexity, but also realizes the distinction between different video streams by the gateway router, which can avoid the limitation of all members receiving the same content in traditional multicast, and provides technical support for personalized learning.

[0118] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0119] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A content reservation and recommendation method for AI online education, characterized by: The content reservation and recommendation method of the AI ​​online education includes: For recorded or delayed live courses, the server sends the original video stream to all student terminals through a preset number of other routers and gateway routers in sequence, while also sending the generated content directory information and preset recommendation information to all student terminals. The gateway router has the functions of both a gateway and a router. Each student can reserve content based on preset recommendation information on the corresponding student terminal as needed; When playing a course, each student terminal detects whether it is a reserved content and adaptively adjusts the playback speed based on the detection result; After each student terminal finishes playing a content point, the server sends questions related to the current content point to the student terminal for the student to answer. The student terminal pauses the playback of the original video stream and determines whether the student's answer accuracy reaches a first threshold. If the first threshold is reached, the original video stream continues to play. If the first threshold is not reached, a request for a detailed explanation video stream of the corresponding content point is sent to the server, and the gateway router stops sending the original video stream to the student terminal that has not reached the first threshold. After receiving the detailed explanation video stream request for each content point, the server sends a first response message to the corresponding student terminal; Each corresponding student terminal parses the received first response information, and when the status of the first response information is acceptance, sends a corresponding detailed video stream request message to the server according to the parsed content; The server then forwards the detailed video stream to the gateway router of the corresponding student terminal through a preset number of other routers; The gateway router of each student terminal sends the detailed video stream to the student terminal that has not reached the first threshold, and caches the original video stream. When the cache space usage of the gateway router is greater than the third threshold, it sends a cache request message to other routers upstream of it.

2. The method for content reservation and recommendation for AI online education according to claim 1, characterized in that: The content directory information includes recorded content directory information and live content directory information; For recorded courses, the complete audio of the course is identified and content points and key content points are marked to form recorded content directory information. The recorded content directory information includes: directory ID, course ID, course type, creation time, content point sequence number, content point name, content point type, content point description, content point start timestamp, content point end timestamp, and whether each content point is a key content point; For delayed live courses, the teacher's course introduction is identified and content points and key content points are marked to form live content directory information. The live content directory information includes: directory ID, course ID, course type, creation time, content point sequence number, content point name, content point type, content point description, and whether each content point is a key content point.

3. The content reservation and recommendation method for AI online education according to claim 2, characterized in that: Each student makes a content reservation based on preset recommendation information on the corresponding student terminal as needed, including: For recorded courses, the preset recommendation information includes at least three recommendation information: recommendation based on the student's own historical reservation content point preference, recommendation based on key content points, and recommendation based on the historical playback content points of other students. For delayed live courses, the preset recommendation information includes at least three recommendation information: recommendation based on the student's own historical reservation content point preference, recommendation based on key content points, and recommendation based on real-time reservation information of other students; Each student can make or not make a reservation for content according to the recommended information as needed; Among them, for recommendations based on real-time reservation information from other student terminals, each student terminal sends the reservation content point information to the teacher terminal and all other student terminals, and the student terminal counts the received reservation content point information, calculates the reservation rate of each content point, and recommends the content point whose reservation rate exceeds the preset recommendation threshold as real-time reservation information for other student terminals.

4. The method for content reservation and recommendation for AI online education according to claim 3, characterized in that: When playing a recorded course or a delayed live course, each student terminal detects whether it is a reserved content and adaptively adjusts the playback speed based on the detection result, including: For recorded courses, the student terminal detects the current playback time point. If the current playback time point is between the start and end timestamps of the scheduled content, the playback is slowed down according to the first preset speed. Otherwise, the playback is accelerated according to the second preset speed. For delayed live courses, the student client dynamically extracts the audio of the currently playing video stream and performs text extraction. Using natural language processing technology, the extracted text and the current student reservation content are vectorized and the difference between the two vectors is calculated. If the difference is less than a second threshold, the video is played at a reduced speed according to the first preset speed. Otherwise, the video is played at a accelerated speed according to the second preset speed. If a student has not reserved any content points, the corresponding student terminal will adaptively adjust the playback speed according to the recommendations for key content points; For recorded courses, when students have not made reservations for all content points and there are no key content points, the corresponding student terminal will adaptively adjust the playback speed based on the recommendations of historical playback content points on other student terminals.

5. The method for content reservation and recommendation of AI online education according to claim 1, characterized in that: The detailed explanation video stream request includes a message ID, a message type, a student ID, a creation time, a course ID, and a content point detailed explanation video stream ID; The first response information includes message ID, message type, student ID, creation time, response status, multicast group address, original video stream port and detailed video stream port; The fields of the detailed video stream request message include message type, checksum, multicast group address, original video stream port and detailed video stream port, wherein the student end adds the original video stream port and detailed video stream port in the parsed first response information to the IGMP v2 message to form a detailed video stream request message.

6. The AI ​​online education content reservation and recommendation method according to claim 1, characterized in that: When the student terminal sends a detailed video stream request message to the server, it is forwarded to the server through the gateway server and other routers. After the gateway server receives the detailed video stream request message, it parses it and records the original video stream port and the detailed video stream port. The gateway router receives the original video stream and the detailed video stream, and distinguishes the original video stream and the detailed video stream through the target port number information of the video stream according to the ACL access control list rules on the gateway router.

7. The method for content reservation and recommendation for AI online education according to claim 1, characterized in that: The gateway router located at the student terminal and n other routers are included between each student terminal and the server, and the n other routers are sequentially represented as the first router P1, the second router P2, ..., the nth router Pn from the gateway router to the server; When caching the original video stream, the gateway router records the time point t1 at which the original video stream is cached and stores it in the mapping table, and continuously monitors whether the cache space usage rate is greater than a third threshold. When the cache space usage rate is greater than the third threshold, the gateway router sends a cache request message to the first router P1 upstream of the gateway router. After receiving the cache request message, the first router P1 decides whether to accept the cache request based on its own cache space; When the first router P1 accepts the cache request, it sends a second response message of acceptance to the gateway router, and the gateway router records the IP address of the first router P1 in the second response message, and records the time point t2 when the first router P1 starts caching the original video stream and stores it in the mapping table. The gateway router itself stops caching the original video stream, and the first router P1 starts caching the original video stream at time point t2. When the first router P1 rejects the cache request, it sends a second response message of rejection to the gateway router, and the gateway router stops sending the detailed video stream to the student terminal that does not reach the first threshold, and resends the original video stream to the student terminal that does not reach the first threshold.

8. The method for content reservation and recommendation of AI online education according to claim 7, characterized in that: When caching the original video stream, the first router P1 continuously monitors whether the cache space usage is greater than a third threshold, and sends a cache request message to the second router P2 adjacent upstream when the usage is greater than the third threshold; After receiving the cache request message, the second router P2 decides whether to accept the cache request based on its own cache space; When the second router P2 accepts the cache request, it simultaneously sends a third response message of acceptance to the first router P1 and the gateway router adjacent to it downstream. The gateway router records the IP address of the second router P2 in the third response message, as well as the time point t3 when the second router P2 starts caching the original video stream, and stores the data in the mapping table. The first router P1 stops caching the original video stream, and the second router P2 starts caching the original video stream at time t3. When the second router P2 rejects the cache request, it only sends a third response message of rejection to its adjacent downstream first router P1, and the first router P1 sends a second response message of rejection to the gateway router. At this time, the gateway router stops sending the detailed video stream to the student terminal that does not reach the first threshold, and resends the original video stream to the student terminal that does not reach the first threshold; When the cache space usage rate of one of the n other routers Px is always less than the third threshold, the gateway router stops sending cache request messages to its other upstream routers, and records the time point t(x+1) when the other router Px starts caching the original video stream and stores it in the mapping table; The mapping table stores the mapping relationship between each time point at which the original video stream starts to be cached and the cache device, and the cache device is each router.

9. The method for content reservation and recommendation for AI online education according to claim 8, characterized in that: The content reservation and recommendation method for AI online education further includes: when the detailed video stream request is finished playing on the student terminal, the corresponding student terminal sends the original video stream request information to the gateway router; After receiving the original video stream request information, the gateway router selects the earliest time point t1 to record the original video stream according to the mapping table, and sends the original video stream with the starting time point t1 to the corresponding student terminal; When the gateway router sends the original video stream to time point t2, according to the mapping table, the gateway router sends the original video stream request information to the cache device first router P1 corresponding to the cache start time point t2; After receiving the original video stream request information, the first router P1 forwards the original video stream starting at time t2 to the corresponding student terminal through the gateway router; When the network router forwards the original video stream to time point t3, according to the mapping table, the network router sends the original video stream request information to the cache device second router P2 corresponding to the cache start time point t3; After receiving the original video stream request information, the second router P2 forwards the original video stream starting at time t3 to the corresponding student terminal through the gateway router; And so on, until other routers Px receive the original video stream request information, they forward it to the corresponding student terminal through the gateway router and send the original video stream with the starting time point t(x+1).

10. The method for content reservation and recommendation of AI online education according to claim 9, characterized in that: The content reservation and recommendation method for AI online education further includes: after each router sends the original video stream to the corresponding student terminal, clearing the cache of the sent original video stream; After the gateway router clears the cache, when it detects that the cache space usage of the gateway router is less than a fourth threshold, it sends a stop caching request message to other routers Px that are caching the latest data of the original video stream; After receiving the stop caching request message from the gateway router, if the other router Px accepts the stop caching, the other router Px stops caching the original video stream and sends a stop caching request response message with a response status of accepted to the gateway router; After receiving the accepted stop caching request response message, the gateway router restarts caching the remaining uncached original video streams.

Images