Learning resource recommendation method, system and device and storage medium

Abstract

The invention belongs to the field of resource recommendation, and particularly relates to a learning resource recommendation method, system and device and a storage medium, the learning time of a user is divided into a fragmented time range and a long-time learning time range, the learning ability of the user is calculated, and according to the learning ability and the association degree of each chapter in a knowledge graph, the learning resource is recommended. When the current time is in the fragmented learning time range, a recommendation emergency value is obtained according to the current learning urgency degree and the knowledge gap risk degree, when the recommendation emergency value is smaller than a preset threshold value, a chapter with a maximum recommendation strategy is matched in the knowledge graph for recommendation, and if the recommendation emergency value is larger than the preset threshold value, the recommendation strategy is matched with the chapter with the maximum recommendation strategy. And if yes, optimizing and recommending chapters corresponding to the urgently needed learning resource type for learning. According to the method and the device, the corresponding learning resource recommendation is carried out in consideration of the time fragments of user learning, and the content which is urgently needed to be learned at present is also considered during recommendation to carry out preferential learning, so that the examination score is improved.

Description

Technical Field

[0001] The present invention belongs to the field of resource recommendation, and particularly relates to a learning resource recommendation method, system, device and storage medium. Background Art

[0002] With the rapid development of the Internet, learning on the Internet has become a relatively mainstream learning method at present. However, there are numerous resources on the Internet, and it is difficult for users to distinguish which resources are suitable for themselves, or it takes a relatively long time to find resources. Therefore, it is very necessary to recommend learning resources.

[0003] In related technologies, learning resource recommendation mainly adopts a classification recommendation method. The classification recommendation method mainly classifies different courses, and users input titles for fuzzy search. Then, according to the results of the fuzzy search, recommendations are made according to the titles associated with the user input, and users select learning resources that may be suitable for themselves according to the title content.

[0004] In view of the above related technologies, when recommending learning resources, only the resources that the user needs to learn are often considered, without considering whether the time the user currently has for learning is sufficient to learn the recommended resources, resulting in the situation that users may not be able to complete learning a chapter in one learning session. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a learning resource recommendation method, system, device and storage medium, which reasonably plan learning resources for recommendation according to the learning time of users, so as to improve the utilization rate and learning efficiency.

[0006] A learning resource recommendation method includes:

[0007] Obtain user personal information, where the personal information includes historical learning accuracy rate, average answering time, education background level, number of completed courses, and total number of course visits;

[0008] Obtain the user learning ability according to the user personal information and the ability learning formula;

[0009] The user learning ability formula is expressed as:

[0010] α = W1*(S / T) + W2log(1 + E) + W3(C / A);

[0011] where S is the historical learning accuracy rate, T is the average answering time, E is the education background level, C is the number of completed courses, A is the total number of course visits, and W1, W2, and W3 are adjustable weight coefficients;

[0012] Based on the historical usage data of the user device, the sliding window algorithm is used to divide time segments in real time to predict the fragmented learning time range and the long-term learning time range when the user is learning;

[0013] Match the knowledge graph according to the user's educational background level;

[0014] Calculate the estimated learning duration of each chapter according to the relevance between the user's learning ability and each chapter in the knowledge graph;

[0015] The calculation method of the estimated learning duration is:

[0016] T i =(L i *(1-β i )+M i *β i ) / α

[0017] Where L i is the basic duration of the chapter, M i is the average migration duration of related chapters, T i is the estimated learning duration, α is the user's learning ability, and β i is;

[0018] When the current learning time is within the fragmented learning time range, obtain the current learning urgency;

[0019] Obtain the knowledge gap risk degree;

[0020] According to the current learning urgency and the knowledge gap risk degree, obtain the recommended emergency value;

[0021] When the recommended emergency value is less than the preset threshold, execute the maximized recommendation strategy, and the maximized recommendation strategy is to aim to learn the most chapters within the fragmented learning time range;

[0022] According to the knowledge graph and the maximized recommendation strategy, match the corresponding chapters as recommended resources and recommend them to the user;

[0023] When the recommended emergency value is greater than the preset threshold, obtain the urgently needed learning resource type according to the current time;

[0024] Match the chapter with the highest similarity to the urgently needed learning resource type in the knowledge graph as the recommended resource and recommend it to the user.

[0025] Optionally, the obtaining of the recommended emergency value according to the current learning urgency and the knowledge gap risk degree includes:

[0026] Construct a time urgency factor according to the current learning urgency;

[0027] The time urgency factor is expressed as:

[0028]

[0029] where η is the time urgency factor, ω is the exam weight coefficient, D c is the deadline, K = 0.2, K is the steepness coefficient of the S-shaped curve, C mastered is the number of mastered chapters, C total is the total number of chapters to be mastered, D now is the current date;

[0030] The knowledge gap risk degree is expressed as:

[0031]

[0032] where ρ i is the knowledge gap risk degree, is the centrality of chapter i in the knowledge graph, s i is the number of days since the last learning, σ = 7, σ is the memory decay period parameter, N attempt is the number of historical learning attempts;

[0033] The recommended emergency value is:

[0034] μ = sigmoid(η * max(ρ i ))

[0035] where μ is the recommended emergency value.

[0036] Optionally, when the current learning time is within the long learning time range, obtain the types of unmastered chapters according to the user's personal information;

[0037] Match the types of unmastered chapters with the knowledge graph to obtain the top n unmastered chapters with the highest similarity as the recommended resource collection;

[0038] Obtain the estimated learning duration of each chapter in the recommended resource collection to be recommended and the difficulty level of the unmastered chapters;

[0039] Filter from the recommended resource collection to be recommended according to the difficulty level and estimated learning duration of the unmastered chapters to obtain the recommended resources;

[0040] Recommend resources to the user according to the recommended resources.

[0041] Optionally, the filtering from the recommended resource collection to be recommended according to the difficulty level and estimated learning duration of the unmastered chapters to obtain the recommended resources includes:

[0042] Sort the to-be-recommended resource collection according to the difficulty levels of the unmastered chapters to obtain a first sorting result;

[0043] Adjust the first sorting result according to the knowledge correlation degree to obtain a second sorting result;

[0044] Use the second sorting result as the first recommended resource;

[0045] Obtain the actual learning time of the user learning the first recommended resource;

[0046] If the actual learning time is greater than a first preset stay time, extract the semantic feature vector of the first recommended resource, and match the top n chapters with the highest similarity as the second recommended resources;

[0047] If the second learning time of the user learning the second recommended resources is less than the preset time, re-push the first recommended resource as the to-be-recommended resource.

[0048] Optionally, if the re-learning time of the user re-learning the first recommended resource is greater than a second preset stay time, obtain the mastered chapters;

[0049] Match the corresponding mastered chapters as the to-be-recommended resources according to the mastering time of the mastered chapters.

[0050] Optionally, the matching the corresponding mastered chapters as the to-be-recommended resources according to the mastering time of the mastered chapters includes:

[0051] Obtain the mastering time weight of the mastered chapters and the chapter weight of the mastered chapters;

[0052] Obtain the total weight according to the mastering time weight of the mastered chapters and the chapter weight of the mastered chapters;

[0053] Sort the mastered chapters according to the total weight to obtain a first weight sorting;

[0054] Set the fatigue learning time;

[0055] Calculate the remaining learning time according to the fatigue learning time, the second preset stay time, and the first preset stay time;

[0056] Adjust the first weight sorting according to the remaining learning time to obtain a second weight sorting;

[0057] Use the first preset number of mastered chapters in the second weight sorting as the to-be-recommended resources.

[0058] Optionally, the adjusting the first sorting result according to the knowledge correlation degree to obtain a second sorting result includes:

[0059] Obtain the knowledge correlation degree of the unmastered chapters in the first sorting result, where the knowledge correlation degree is the knowledge learning order between the unmastered chapters;

[0060] If there is no knowledge correlation degree in the first sorting result, then use the first sorting result as the second sorting result;

[0061] If there is a knowledge correlation degree in the first sorting result, then adjust the first sorting result according to the knowledge correlation degree to obtain the second sorting result.

[0062] A learning resource recommendation system, including:

[0063] A first acquisition module, used to obtain user personal information, where the personal information includes historical learning accuracy rate, average answering time, education background level, number of completed courses, and total number of course visits;

[0064] A first calculation module, used to obtain the user learning ability according to the user personal information and the ability learning formula;

[0065] The user learning ability formula is expressed as:

[0066] α = W1*(S / T) + W2log(1 + E) + W3(C / A);

[0067] Where S is the historical learning accuracy rate, T is the average answering time, E is the education background level, C is the number of completed courses, A is the total number of course visits, and W1, W2, and W3 are adjustable weight coefficients;

[0068] A prediction module, used to divide time segments in real time based on the historical usage data of the user device by using a sliding window algorithm, and predict the fragmented time range and long-time learning time range when the user is learning;

[0069] A first matching module, used to match the knowledge graph according to the user's education background level;

[0070] A second calculation module, used to calculate the estimated learning duration of each chapter according to the user learning ability and the correlation degree of each chapter in the knowledge graph;

[0071] The calculation method of the estimated learning duration is:

[0072] T i =(L i *(1 - β i ) + M i *β i ) / α

[0073] Where L i is the basic duration of the chapter, M i is the average transfer duration of the associated chapters, Ti For the estimated learning duration, α is the user's learning ability, β i is the chapter relevance, and i is the i-th chapter;

[0074] A second acquisition module, configured to acquire the current learning urgency when the current learning time is within the fragmented learning time range;

[0075] A third acquisition module, configured to acquire the knowledge gap risk degree;

[0076] A third calculation module, configured to obtain a recommended emergency value according to the current learning urgency and the knowledge gap risk degree;

[0077] A first recommendation module, configured to execute a maximization recommendation strategy when the recommended emergency value is less than a preset threshold, and the maximization recommendation strategy is to aim at learning the most chapters within the fragmented learning time range;

[0078] A second matching module, configured to match corresponding chapters as recommended resources and recommend them to the user according to the knowledge graph and the maximization recommendation strategy;

[0079] A third matching module, configured to obtain the urgently needed learning resource type according to the current time when the recommended emergency value is greater than a preset value;

[0080] A second recommendation module, configured to match the chapters with the highest similarity to the urgently needed learning resource type from the knowledge graph as recommended resources and recommend them to the user.

[0081] A terminal device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, a learning resource recommendation method is adopted.

[0082] A computer-readable storage medium stores a computer program. When the computer program is loaded and executed by a processor, a learning resource recommendation method is adopted.

[0083] The beneficial effects of the present invention are:

[0084] 1. Divide the user's learning time into fragmented time ranges and long-term learning time ranges, then calculate the user's learning ability. Based on the learning ability and the correlation of each chapter in the knowledge graph, obtain the estimated learning duration of each chapter. Then, when the current time is within the fragmented learning time range, obtain the recommended emergency value according to the current learning urgency and knowledge gap risk. When the recommended emergency value is less than the preset threshold, match the chapter that maximizes the recommended strategy from the knowledge graph for recommendation, that is, learn the most chapters within a certain period of time. If it is greater than the preset threshold, optimize the recommendation to learn the chapters corresponding to the urgently needed learning resource types. This application takes into account the fragmented time of the user's learning and makes corresponding learning resource recommendations. Moreover, when recommending, it also considers the content that needs to be learned urgently for priority learning, thereby improving the test scores.

[0085] 2. Within the long-term learning time range, not only consider the difficulty of the chapters, but also consider the handling method when the user has not learned for a long time during learning, that is, recommend learning resources of the same type, and also consider the situation when the user relaxes learning if the learning time is too long. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 It is a flowchart of a learning resource recommendation method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0087] A learning resource recommendation method, as Figure 1 shown, includes:

[0088] S1. Obtain the user's personal information, where the personal information includes historical learning accuracy rate, average answering time, educational background level, number of completed courses, and total course access times.

[0089] S2. Obtain the user's learning ability according to the user's personal information and the learning ability formula;

[0090] The user learning ability formula is expressed as:

[0091] α = W1*(S / T) + W2log(1 + E) + W3(C / A);

[0092] where S is the historical learning accuracy rate, T is the average answering time, E is the educational background level, C is the number of completed courses, A is the total course access times, and W1, W2, and W3 are adjustable weight coefficients;

[0093] Specifically, the user's learning ability is mainly reflected in the accuracy rate when the user is learning, the amount of courses completed in learning, and the educational level of the current learning resources.

[0094] The educational background levels are divided according to different learning stages such as preschool education, primary school, junior high school, and senior high school. Benchmark values are set for the educational levels at each stage. For example, the benchmark value for preschool education is 0.5, for primary school is 1, for junior high school is 1.5, and for senior high school is 2. Adjustment coefficients can also be set according to the educational quality of the school. For example, if the educational quality of a senior high school user's school is relatively good, the coefficient is set to 1.1, and the educational background level of this user is calculated as 2.2.

[0095] S3. Based on the historical usage data of the user device, use the sliding window algorithm to divide time segments in real time, and predict the fragmented learning time range and long-term learning time range when the user is learning;

[0096] Specifically, based on the historical usage data of the user device, obtain the time when the user is continuously in the learning state and the interval time between two learning activities. Set the sliding size, which is 30 minutes in this embodiment, and set the sliding step size to 1. Slide the user's learning time throughout the day. The time greater than the sliding window is defined as the long-term learning time range, otherwise it is the fragmented learning time range.

[0097] S4. Match the knowledge graph according to the user's educational background level.

[0098] Specifically, different educational background levels require different knowledge to be learned. Therefore, according to the user's educational background level, match the corresponding knowledge chapters to be learned and generate a knowledge graph. It can be generated in the form of the number of chapters according to the topological structure of the curriculum syllabus. The content included in the knowledge graph includes both the content of the book and the content on the Internet.

[0099] S5. Calculate the estimated learning duration for each chapter according to the user's learning ability and the correlation degree of each chapter in the knowledge graph;

[0100] The calculation method of the estimated learning duration is:

[0101] T i =(L i *(1-β i )+M i *β i ) / α

[0102] Where L i is the basic duration of the chapter, M i is the average migration duration of related chapters, T i is the estimated learning duration, α is the user's learning ability, β i is the chapter correlation degree, and i is the i-th chapter;

[0103] Specifically, when learning two or more pieces of knowledge with chapter relevance, there is partial correlation. For such chapters, the learning time needs to be adjusted accordingly, that is, the average transfer duration of related chapters * chapter relevance.

[0104] S6. When the current learning time is within the fragmented learning time range, obtain the current learning urgency.

[0105] Specifically, the current learning urgency refers to the urgency at the current time from a certain deadline, such as the time until the mid-term exam.

[0106] S7. Obtain the knowledge gap risk degree.

[0107] Specifically, the knowledge gap risk degree is the importance of the unmastered chapters in a certain goal (mid-term exam).

[0108] S8. Obtain the recommended emergency value according to the current learning urgency and the knowledge gap risk degree;

[0109] Obtaining the recommended emergency value according to the current learning urgency and the knowledge gap risk degree includes:

[0110] Construct a time urgency factor according to the current learning urgency;

[0111] The time urgency factor is expressed as:

[0112]

[0113] Among them, η is the time urgency factor, ω is the exam weight coefficient, D c is the deadline, K = 0.2, K is the steepness coefficient of the S-shaped curve, C mastered is the number of mastered chapters, C total is the total number of chapters to be mastered, D now is the current date;

[0114] The knowledge gap risk degree is expressed as:

[0115]

[0116] Among them, ρ i is the knowledge gap risk degree, is the centrality of chapter i in the knowledge graph, s i is the number of days since the last learning, σ = 7, σ is the memory decay cycle parameter, N attempt is the number of historical learning attempts;

[0117] The recommended emergency value is:

[0118] μ = sigmoid(η * max(ρ i))

[0119] Among them, μ is the recommended emergency value.

[0120] Specifically, the recommended emergency value is calculated by considering the current learning urgency and the risk degree of knowledge gap. The larger the value, the more urgent it is.

[0121] S9. When the recommended emergency value is less than the preset threshold, execute the maximized recommendation strategy. The maximized recommendation strategy aims to learn the most chapters within the fragmented learning time range.

[0122] S10. According to the knowledge graph and the maximized recommendation strategy, match the corresponding chapters as recommended resources and recommend them to the user.

[0123] Specifically, when the recommended emergency value is less than the preset threshold, it means that there are no particularly urgent learning resources that need to be learned preferentially.

[0124] S11. When the recommended emergency value is greater than the preset threshold, obtain the urgently needed learning resource type according to the current time;

[0125] Specifically, according to the current time and the user's preset exam time, obtain the next exam type, including mid-term exams, final exams, and monthly exams, match the corresponding exam scope, and then match within the exam scope and the knowledge graph to obtain the urgently needed learning resource type.

[0126] S12. Match the chapter with the highest similarity to the urgently needed learning resource type in the knowledge graph as the recommended resource and recommend it to the user.

[0127] Specifically, the learning resources can be obtained through similarity calculation:

[0128]

[0129] Among them, C k represents the k-th cluster. The learners in the k-th cluster can be described by the mean or centroid. |S(i|C k )| represents the number of learners who click on item i in the learner cluster C k . |S(i m |C k )∩S(i n |C k )| represents the number of learners who click on both i k and i m in the learner cluster C n . S(x) represents the set of resources browsed by learner x. S(i m ,L) represents the L most similar recommended results to the science and education resource i m .

[0130] It can also be matched through the keywords of each chapter. For example, Chapter A has 5 keywords and Chapter B has 6 keywords, and 3 of them are the same, then the similarity is considered to be 3 / 6.

[0131] When the current learning time is within the long learning time range, obtain the types of unmastered chapters according to the user's personal information.

[0132] Specifically, the types of unmastered chapters are the types of chapters that the user has not learned or has not learned after learning.

[0133] Match the types of unmastered chapters with the knowledge graph to obtain the top n unmastered chapters with the highest similarity as the recommended resource collection.

[0134] Specifically, the user may have multiple types of unmastered chapters. At this time, the order can be selected according to the order of the curriculum outline, or the similarity calculation can be performed on all types of unmastered chapters, and the recommended resources corresponding to each type of unmastered chapter can be used as the recommended resource collection.

[0135] Obtain the estimated learning duration of each chapter in the recommended resource collection to be recommended and the difficulty of the unmastered chapters;

[0136] Specifically, the difficulty of the unmastered chapters can be obtained according to the curriculum outline, or can be set in advance according to expert experience and then obtained from the database.

[0137] Screen the recommended resource collection to be recommended according to the difficulty of the unmastered chapters and the estimated learning duration to obtain the recommended resources.

[0138] Specifically, the user has many unmastered chapters and different difficulties, but the duration of the long learning time is also different. For example, the user has two long learning time ranges every day, one is from 7:00 to 7:40 in the morning, and the other is from 6:30 to 7:30 in the evening. For some more difficult chapters, there may not be enough time to learn in the morning. Therefore, screening is required to leave the resources that can be learned at the current time to obtain the recommended resources.

[0139] Recommend resources to the user according to the recommended resources.

[0140] Screen the recommended resource collection to be recommended according to the difficulty of the unmastered chapters and the estimated learning duration to obtain the recommended resources including:

[0141] Sort the recommended resource collection to be recommended according to the difficulty of the unmastered chapters to obtain the first sorting result;

[0142] Specifically, since it is the content that can be learned sufficiently during a long study time, when there is enough time, it is difficult to optimize the arrangement of learning more difficult chapters. In the subsequent learning time, the difficulty gradually decreases, and the remaining time can be used for review. Or, after learning one chapter in sequence every day, the remaining time can be converted into fragmented time to learn the chapters corresponding to the fragmented time, so as to improve the time utilization rate.

[0143] Adjust the first sorting result according to the knowledge correlation degree to obtain the second sorting result;

[0144] Specifically, although the sorting is carried out according to the difficulty level, for some chapters, although they are relatively simple, they are the basis for other more difficult chapters. In this case, it is necessary to learn the basic chapters first and adjust the order of this chapter to be in front of the more difficult chapters.

[0145] Take the second sorting result as the first recommended resource.

[0146] Obtain the actual learning time of the user learning the first recommended resource;

[0147] If the actual learning time is greater than the first preset stay time, extract the semantic feature vector of the first recommended resource and match the top n chapters with the highest similarity as the second recommended resource;

[0148] Specifically, when the user learns according to the first recommended resource, for each chapter, combined with the user's learning ability, there is a corresponding estimated learning time. If the user's actual learning time is greater than the first preset stay time, and the first preset stay time is the user's corresponding estimated learning time * the first adjustment coefficient, such as 1.2. That is to say, if the user still doesn't understand after being given enough time, it is determined that the user has fallen into a bottleneck. Therefore, extract the semantic feature vector of the first recommended resource and match the top n chapters with the highest similarity as the second recommended resource. The feature vector can be obtained through models such as BERT. The purpose is to give the user inspiration through other types of chapters.

[0149] If the second learning time of the user learning the second recommended resource is less than the preset time, re-push the first recommended resource as the resource to be recommended.

[0150] Specifically, if the second learning time of the user learning the second recommended resource is less than the preset time, it means that the user has understood the second recommended resource. Therefore, re-push the learned first recommended resource so that the user can learn the first recommended resource.

[0151] If the re-learning time of the user re-learning the first recommended resource is greater than the second preset stay time, obtain the mastered chapters.

[0152] Specifically, if the user still spends a large amount of time learning the second recommended resource of the first recommended resource type, the second preset stay time can be obtained by multiplying the estimated time by the corresponding second adjustment coefficient, and the second adjustment coefficient is less than the first adjustment coefficient.

[0153] Match the corresponding mastered chapters as the resources to be recommended according to the mastery time of the mastered chapters.

[0154] Specifically, when the user fails to master the first recommended resource twice, it is no longer suitable to continue learning this resource at present. And since a relatively long time has been spent thinking, it is possible to appropriately learn something that is already relatively easy to relax. The mastery time is the time from when the user learned this chapter to the present.

[0155] Matching the corresponding mastered chapters as the resources to be recommended according to the mastery time of the mastered chapters includes:

[0156] Obtain the mastery time weight of the mastered chapters and the chapter weight of the mastered chapters.

[0157] Specifically, different weights are set according to different time spans. For example, one month is one weight, and two months is one weight.

[0158] The chapter weight can be determined according to the score proportion of this chapter in the exam.

[0159] Obtain the total weight according to the mastery time weight of the mastered chapters and the chapter weight of the mastered chapters;

[0160] Specifically, if the total weights are the same, then the one with the larger chapter weight is given priority. If the chapter weights are also the same, then a random recommendation is made.

[0161] Sort the mastered chapters according to the total weight to obtain the first weight sorting;

[0162] Set the fatigue learning time.

[0163] Specifically, the fatigue learning time is the longest time for the user to study at one time. If it exceeds this time, the user's learning efficiency will decrease significantly. The fatigue learning time can be set by oneself or determined according to the user's past learning efficiency.

[0164] Calculate the remaining learning time according to the fatigue learning time, the second preset stay time, and the first preset stay time.

[0165] Adjust the first weight sorting according to the remaining learning time to obtain the second weight sorting.

[0166] Specifically, the remaining learning time = fatigue learning time - second preset stay time - first preset stay time.

[0167] Specifically, for the chapters sorted by the first weight, some of the learning time may already be greater than the remaining learning time. Therefore, the chapters that need less review time than the remaining learning time in the mastered chapters will be re-sorted in ascending order of review time.

[0168] The time required to review the mastered chapters can be obtained by multiplying the estimated learning time when learning the chapter by the review weight. The review weight can be adjusted in combination with the difficulty level, or a fixed adjustment coefficient can be set, and the adjustment coefficient is less than 1.

[0169] Take the first preset number of mastered chapters sorted by the second weight as the resources to be recommended.

[0170] Adjust the first sorting result according to the knowledge correlation degree to obtain the second sorting result, including:

[0171] Obtain the knowledge correlation degree of the unmastered chapters in the first sorting result. The knowledge correlation degree is the knowledge learning order between the unmastered chapters;

[0172] If there is no knowledge correlation degree in the first sorting result, then take the first sorting result as the second sorting result;

[0173] If there is a knowledge correlation degree in the first sorting result, then adjust the first sorting result according to the knowledge correlation degree to obtain the second sorting result.

[0174] Specifically, the knowledge learning order between the unmastered chapters can be obtained through the keywords of the unmastered chapters, or through the book outline. Obtaining through keywords can be, for example, if the keyword of chapter A is matrix and the keyword of chapter B is matrix calculation, then chapter A needs to optimize the learning of chapter B.

[0175] A learning resource recommendation system, including:

[0176] The first acquisition module is used to acquire the user's personal information, and the personal information includes historical learning accuracy rate, average answering time, education background level, number of completed courses, and total course access times;

[0177] The first calculation module is used to obtain the user's learning ability according to the user's personal information and the ability learning formula;

[0178] The user learning ability formula is expressed as:

[0179] α = W1*(S / T) + W2log(1 + E) + W3(C / A);

[0180] Where S is the historical learning accuracy rate, T is the average answering time, E is the education background level, C is the number of completed courses, A is the total course access times, and W1, W2, and W3 are adjustable weight coefficients;

[0181] A prediction module, which is used to divide time segments in real time by using a sliding window algorithm based on the historical usage data of the user device, and predict the fragmented time range and long-time learning time range when the user is learning;

[0182] A first matching module, which is used to match a knowledge graph according to the educational background level of the user;

[0183] A second calculation module, which is used to calculate the estimated learning duration of each chapter according to the user's learning ability and the correlation degree of each chapter in the knowledge graph;

[0184] The calculation method of the estimated learning duration is:

[0185] T i =(L i *(1-β i )+M i *β i ) / α

[0186] where L i is the basic duration of the chapter, M i is the average migration duration of related chapters, T i is the estimated learning duration, α is the user's learning ability, β i is the chapter correlation degree, and i is the i-th chapter;

[0187] A second acquisition module, which is used to acquire the current learning urgency when the current learning time is within the fragmented learning time range;

[0188] A third acquisition module, which is used to acquire the knowledge gap risk degree;

[0189] A third calculation module, which is used to obtain a recommended emergency value according to the current learning urgency and the knowledge gap risk degree;

[0190] A first recommendation module, which is used to execute a maximization recommendation strategy when the recommended emergency value is less than a preset threshold, and the maximization recommendation strategy is to aim at learning the most chapters within the fragmented learning time range;

[0191] A second matching module, which is used to match the corresponding chapters as recommended resources and recommend them to the user according to the knowledge graph and the maximization recommendation strategy;

[0192] A third matching module, which is used to obtain the type of urgently needed learning resources according to the current time when the recommended emergency value is greater than the preset value;

[0193] A second recommendation module, which is used to match the chapters with the highest similarity to the type of urgently needed learning resources from the knowledge graph as recommended resources and recommend them to the user.

[0194] An embodiment of the present application also discloses a terminal device, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, a learning resource recommendation method is adopted.

[0195] Among them, the terminal device can be a computer device such as a desktop computer, a laptop computer, or a cloud server. Moreover, the terminal device includes, but is not limited to, a processor and a memory. For example, the terminal device may further include an input / output device, a network access device, and a bus, etc.

[0196] Among them, the processor can adopt a central processing unit (CPU). Of course, according to the actual usage situation, other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general-purpose processor can adopt a microprocessor or any conventional processor, etc. The present application does not make any restrictions in this regard.

[0197] Among them, the memory can be an internal storage unit of the terminal device. For example, the hard disk or memory of the terminal device, or it can also be an external storage device of the terminal device. For example, a plug-in hard disk, a smart media card (SMC), a secure digital card (SD), or a flash card (FC), etc. equipped on the terminal device. Moreover, the memory can also be a combination of the internal storage unit and the external storage device of the terminal device. The memory is used to store the computer program and other programs and data required by the terminal device. The memory can also be used to temporarily store the data that has been output or will be output. The present application does not make any restrictions in this regard.

[0198] Among them, through this terminal device, a learning resource recommendation method in the above embodiment is stored in the memory of the terminal device, and is loaded and executed on the processor of the terminal device, which is convenient for use.

[0199] An embodiment of the present application also discloses a computer-readable storage medium. Moreover, the computer-readable storage medium stores a computer program. When the computer program is executed by the processor, a learning resource recommendation method in the above embodiment is adopted.

[0200] Among them, the computer program can be stored in a computer-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some middleware form, etc. The computer-readable medium includes any entity or device, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the computer-readable medium includes, but is not limited to, the above components.

[0201] Among them, through this computer-readable storage medium, the learning resource recommendation method in the above embodiments is stored in the computer-readable storage medium, and is loaded and executed on the processor to facilitate the storage and application of the above method.

[0202] Those of ordinary skill in the art should understand that: the discussion of any of the above embodiments is only exemplary and is not intended to imply that the protection scope of the present application is limited to these examples; under the idea of the present application, the technical features in the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations in different aspects of one or more embodiments in the present application as above, which are not provided in detail for the sake of brevity.

[0203] One or more embodiments of the present application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present application shall be included within the protection scope of the present application.

Claims

1. A learning resource recommendation method, characterized in that, Including: Obtain the user's personal information, which includes historical learning accuracy rate, average answering time, education background level, number of completed courses, and total number of course visits; Obtain the user's learning ability according to the user's personal information and the ability learning formula; The user's learning ability formula is expressed as: α = W1*(S / T) + W2log(1 + E) + W3(C / A); Where S is the historical learning accuracy rate, T is the average answering time, E is the education background level, C is the number of completed courses, A is the total number of course visits, and W1, W2, and W3 are adjustable weight coefficients; Based on the historical usage data of the user device, use the sliding window algorithm to divide time segments in real time, and predict the fragmented learning time range and long-term learning time range when the user is learning; Match the knowledge graph according to the user's education background level; Calculate the estimated learning duration of each chapter according to the user's learning ability and the relevance of each chapter in the knowledge graph; The calculation method of the estimated learning duration is: T i = (L i * (1 - β i ) + M i * β i ) / α where L i is the basic duration of a chapter, M i is the average migration duration of related chapters, T i is the estimated learning duration, α is the user's learning ability, β i is the chapter correlation degree, and i is the i-th chapter; When the current learning time is within the fragmented learning time range, obtain the current learning urgency; Obtain the knowledge gap risk degree; Obtain the recommended emergency value according to the current learning urgency and the knowledge gap risk degree; When the recommended emergency value is less than the preset threshold, execute the maximization recommendation strategy, and the maximization recommendation strategy is to aim at learning the most chapters within the fragmented learning time range; Match the corresponding chapters as recommended resources according to the knowledge graph and the maximization recommendation strategy and recommend them to the user; When the recommended emergency value is greater than the preset threshold, obtain the urgently needed learning resource type according to the current time; Match the chapter with the highest similarity to the urgently needed learning resource type from the knowledge graph as the recommended resource and recommend it to the user.

2. The learning resource recommendation method according to claim 1, characterized in that The obtaining the recommended emergency value according to the current learning urgency and the knowledge gap risk degree includes: Construct a time urgency factor according to the current learning urgency; The time urgency factor is expressed as: Among them, η is the time urgency factor, ω is the exam weight coefficient, D c is the deadline, K = 0.2, K is the steepness coefficient of the S-shaped curve, C mastered is the number of mastered chapters, C total is the total number of chapters to be mastered, D now is the current date; The knowledge gap risk degree is expressed as: Among them, ρ i is the notch risk degree, is the centrality of chapter i in the knowledge graph, s i is the number of days since the last learning, σ = 7, σ is the memory decay cycle parameter, N attempt is the number of historical learning attempts; The recommended emergency value is: μ=sigmoid(η*max(ρ i )) Where μ is the recommended emergency value.

3. The learning resource recommendation method according to claim 1, wherein Including: When the current learning time is within the long-term learning time range, obtain the types of unmastered chapters according to the user's personal information; Match the types of unmastered chapters with the knowledge graph to obtain the top n unmastered chapters with the highest similarity as the recommended resource collection; Obtain the estimated learning duration of each chapter in the to-be-recommended resource collection and the difficulty of the unmastered chapters; Filter from the to-be-recommended resource collection according to the difficulty and estimated learning duration of the unmastered chapters to obtain the to-be-recommended resources; Recommend resources to the user according to the to-be-recommended resources.

4. The learning resource recommendation method according to claim 3, wherein The filtering from the to-be-recommended resource collection according to the difficulty and estimated learning duration of the unmastered chapters to obtain the to-be-recommended resources includes: Sort the to-be-recommended resource collection according to the difficulty of the unmastered chapters to obtain the first sorting result; Adjust the first sorting result according to the knowledge relevance to obtain the second sorting result; Use the second sorting result as the first recommended resource; Obtain the actual learning time of the user for the first recommended resource; If the actual learning time is greater than the first preset stay time, extract the semantic feature vector of the first recommended resource and match the top n chapters with the highest similarity as the second recommended resources; If the second learning time of the user for the second recommended resource is less than the preset time, re-push the first recommended resource as the resource to be recommended.

5. The learning resource recommendation method according to claim 4, characterized in that, Include: If the re-learning time of the user for the first recommended resource is greater than the second preset stay time, obtain the mastered chapters; Match the corresponding mastered chapters as the resources to be recommended according to the mastery time of the mastered chapters.

6. The learning resource recommendation method according to claim 5, characterized in that The matching of the corresponding mastered chapters as the resources to be recommended according to the mastery time of the mastered chapters includes: Obtain the mastery time weight of the mastered chapters and the chapter weight of the mastered chapters; Obtain the total weight according to the mastery time weight of the mastered chapters and the chapter weight of the mastered chapters; Sort the mastered chapters according to the total weight to obtain the first weight sorting; Set the fatigue learning time; Calculate the remaining learning time according to the fatigue learning time, the second preset stay time and the first preset stay time; Adjust the first weight sorting according to the remaining learning time to obtain the second weight sorting; Take the first preset number of mastered chapters in the second weight sorting as the resources to be recommended.

7. The learning resource recommendation method according to claim 4, characterized in that The adjustment of the first sorting result according to the knowledge association degree to obtain the second sorting result includes: Obtain the knowledge association degree of the unmastered chapters in the first sorting result, and the knowledge association degree is the knowledge learning order between the unmastered chapters; If there is no knowledge association degree in the first sorting result, take the first sorting result as the second sorting result; If there is a knowledge association degree in the first sorting result, adjust the first sorting result according to the knowledge association degree to obtain the second sorting result.

8. A learning resource recommendation system, characterized in that, Include: The first acquisition module is used to acquire user personal information, and the personal information includes historical learning accuracy rate, average answering time, education background level, number of completed courses and total number of course visits; The first calculation module is used to obtain the user learning ability according to the user personal information and the ability learning formula; The user learning ability formula is expressed as: α = W1*(S / T) + W2log(1 + E) + W3(C / A); Where S is the historical learning accuracy rate, T is the average answering time, E is the education background level, C is the number of completed courses, A is the total number of course visits, and W1, W2 and W3 are adjustable weight coefficients; The prediction module is used to divide time segments in real time based on the historical usage data of the user device by using a sliding window algorithm, and predict the fragmented time range and long-time learning time range when the user is learning; The first matching module is used to match the knowledge graph according to the user's education background level; The second calculation module is used to calculate the estimated learning duration of each chapter according to the user learning ability and the association degree of each chapter in the knowledge graph; The calculation method of the estimated learning duration is: T i = (L i * (1 - β i ) + M i * β i ) / α Among them, L i is the basic duration of the chapter, M i is the average migration duration of related chapters, T i is the estimated learning duration, α is the user's learning ability, β i is the chapter correlation degree, and i is the i-th chapter; The second acquisition module is used to obtain the current learning urgency when the current learning time is within the fragmented learning time range; A third acquisition module, configured to acquire the knowledge gap risk degree; A third calculation module, configured to obtain a recommended urgency value according to the current learning urgency and the knowledge gap risk degree; A first recommendation module, configured to execute a maximization recommendation strategy when the recommended urgency value is less than a preset threshold, where the maximization recommendation strategy aims to learn the most chapters within the fragmented learning time range; A second matching module, configured to match corresponding chapters as recommended resources and recommend them to the user according to the knowledge graph and the maximization recommendation strategy; A third matching module, configured to obtain the urgently needed learning resource type according to the current time when the recommended urgency value is greater than a preset value; A second recommendation module, configured to match the chapter with the highest similarity to the urgently needed learning resource type from the knowledge graph as the recommended resource and recommend it to the user.

9. A terminal device, comprising a memory and a processor, characterized in that The memory stores a computer program capable of running on a processor. When the processor loads and executes the computer program, the recommendation method described in any one of claims 1 to 7 is adopted.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by the processor, the recommendation method described in any one of claims 1 to 7 is adopted.

Images