Personalized learning resource recommendation method and system based on improved pigeon inspired algorithm

By improving the pigeon flock algorithm, combining the multi-dimensional matching fitness function and landmark guidance and chaos mapping strategies, the recommendation of personalized learning resources is optimized, and the problems of low recommendation efficiency and insufficient matching in the existing technology are solved, achieving more efficient and stable Personalized learning experience.

CN119025756AActive Publication Date: 2024-11-26NANJING XIAOZHUANG UNIV

Patent Information

Application Number
CN202411125891.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-16
Publication Date
2024-11-26
Estimated Expiration
2044-08-16

AI Technical Summary

Technical Problem

The existing personalized learning resource recommendation algorithm has limitations in accurately capturing the real needs of learners. The recommendation process is low, the accuracy of recommendation results is insufficient, the degree of personalized adaptation of learning paths needs to be strengthened, and it is easy to fall into local optimality, affecting learners. Feature and resource feature matching degree.

Method used

Using the improved pigeon flock algorithm, randomly initialize the pigeon flock position and speed, iteratively update individual positions, introduce landmark guidance and chaos mapping dynamic step size local search variation strategies, combine multi-dimensional matching fitness function, optimize learning resource recommendations, and enhance the global algorithm Explore and local development capabilities to avoid premature maturity and convergence.

Benefits of technology

It improves the accuracy and matching of personalized learning resource recommendations, provides a richer, diverse and personalized learning experience, and enhances the stability and convergence performance of the recommendation system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025756A_ABST
    Figure CN119025756A_ABST
Patent Text Reader

Abstract

The invention discloses a personalized learning resource recommendation method and system based on an improved pigeon inspired algorithm. The method comprises the steps of obtaining personalized features and learning requirements of a learner; inputting the personalized features and the learning requirements into a pre-trained learning resource recommendation model, and constructing a multi-dimensional matching fitness function between the learner and personalized learning resources by the learning resource recommendation model according to the personalized features and the learning requirements of the learner, and solving the multi-dimensional matching fitness function through an improved pigeon inspired algorithm and outputting an optimal learning resource sequence. According to the personalized learning resource recommendation method and system based on the improved pigeon inspired algorithm, a learning resource sequence conforming to the ability level and interest preference of a learner can be generated according to personalized features and learning requirements of the learner, and richer, diversified and personalized learning experience is provided for the learner.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to a personalized learning resource recommendation method and system based on an improved pigeon flock algorithm, belonging to the technical field of smart education. Background Art

[0002] In today's era, the rapid development of artificial intelligence technology and the deep integration of educational applications are reshaping the learning ecosystem at an unprecedented speed. A personalized learning system framework that is guided by teacher experience and learners' active exploration, supported by a finely constructed domain-specific knowledge model, is gradually maturing and improving. This type of system not only effectively solves the problems of geographical limitations, fixed time and limited resources in the traditional education model, but also stands out in the field of smart education through its convenience, instant interaction and personalized learning experience, becoming an important force in promoting educational innovation.

[0003] The realization of personalized learning lies in the solid construction of three pillars: deep computing ability of learning situations, accurate understanding ability of learning subject needs, and highly customized learning services. Faced with the different cognitive abilities of learners, how to accurately screen, reasonably combine and dynamically generate learning resource sequences that meet the unique abilities and needs of each learner from a large amount of learning resources has become the key to improving the efficiency of personalized learning and realizing personalized education. Therefore, the development of intelligent, dynamic and personalized learning resource recommendation mechanisms has become a research hotspot and frontier direction in the field of smart education. However, there are many challenges in the promotion of personalized learning. Learners' personalized characteristics are highly dynamic and unpredictable, and it is difficult to fully describe them through simple quantification. At the same time, the extensiveness and diversity (i.e., massive and heterogeneous) characteristics of personalized learning resources have brought significant difficulties to the recommendation of learning resources. Specifically, the current recommendation algorithm has limitations in accurately capturing the real needs of learners, the efficiency of the recommendation process needs to be improved, the accuracy of the recommendation results does not meet expectations, and the degree of personalized adaptation of the learning path still needs to be strengthened. These problems constitute key obstacles in the process of achieving personalized learning goals, and require interdisciplinary technological innovation and optimization strategies to overcome them one by one.

[0004] In the context of the rapid development of "artificial intelligence + education", researchers have built multi-objective optimization models and used swarm intelligence optimization algorithms, such as particle swarm optimization algorithm ([1] Yang Chao. Learning resource recommendation method based on particle swarm optimization algorithm [J]. Computer Applications, 2014, 34(05): 1350-1353. [2] Li Haojun, Liu Zhongfeng, Li Sai, et al. Personalized network learning resource recommendation method based on improved binary particle swarm algorithm [J]. Systems Science and Mathematics, 2017, 37(08): 1770-1779.) and other classic methods in order to tailor learning plans for each student. By continuously optimizing algorithm parameters and improving search strategies, researchers strive to improve the accuracy of personalized learning resource recommendations and the optimization effect of learning paths, thereby effectively promoting the improvement of personalized learning efficiency. However, although these technological innovations have brought considerable progress, evolutionary computing technology still faces challenges in overcoming premature convergence, reducing repeated searches, and balancing local exploration and global development capabilities, and further technological breakthroughs are needed. In particular, as the number of iterations increases, the population diversity is gradually lost, which makes it easier for the algorithm to fall into the dilemma of local optimality, thereby affecting the matching degree between learner characteristics and learning resource characteristics.

[0005] It can be seen that in order to generate a learning resource sequence that meets the learners' ability level and interest preferences according to their personalized characteristics and learning needs, there is an urgent need for a personalized learning resource recommendation method and system based on the improved pigeon flock algorithm. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a personalized learning resource recommendation method and system based on the improved pigeon flock algorithm, which can generate a learning resource sequence that meets the learner's ability level and interest preference according to the learner's personalized characteristics and learning needs, and provide learners with a richer, more diverse and personalized learning experience. At the same time, in the constructed learning resource recommendation model, innovative multi-dimensional evaluation indicators are defined, including the matching degree between learner characteristics and learning resource characteristics, learning time cost, and knowledge point coverage and distribution balance.

[0007] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0008] In a first aspect, the present invention provides a personalized learning resource recommendation method based on an improved pigeon flock algorithm, comprising:

[0009] Obtain learners' personalized characteristics and learning needs;

[0010] The personalized characteristics and learning needs are input into a pre-trained learning resource recommendation model, the learning resource recommendation model constructs a multi-dimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, solves the multi-dimensional matching fitness function through an improved pigeon flock algorithm and outputs an optimal learning resource sequence;

[0011] The improved pigeon flock algorithm comprises:

[0012] Randomly initialize the position and speed of the pigeon flock;

[0013] Iterate the first iteration step until the number of iterations of the first iteration step reaches a preset first maximum number of iterations, then execute the second iteration step, and output the optimal learning resource sequence by the second iteration step;

[0014] The second iteration step comprises:

[0015] Update the individual positions of the population based on landmark guidance;

[0016] Update discovery probability;

[0017] Determine whether to disturb the optimal individual. If so, implement the dynamic step size local search mutation strategy based on chaotic mapping to disturb the optimal individual. Determine whether to retain the disturbed individual. Otherwise, determine whether the values ​​of each matching function are lower than the threshold or whether the second iteration step reaches the preset second maximum number of iterations. If any of the conditions are met, output the optimal learning resource sequence, otherwise re-execute the second iteration step.

[0018] Furthermore, the iterative step includes:

[0019] Update the individual positions of the population based on landmark guidance;

[0020] Update discovery probability;

[0021] Determine whether to disturb the optimal individual. If so, implement the dynamic step size local search mutation strategy based on chaotic mapping to disturb the optimal individual. Determine whether to retain the disturbed individual. Otherwise, determine whether the values ​​of each matching function are lower than the threshold or whether the hybrid optimization stage of the algorithm has reached the maximum number of iterations. If any of the conditions are met, output the optimal learning resource, otherwise re-execute the iterative step.

[0022] Furthermore, the multidimensional matching fitness function includes five secondary fitness functions, wherein:

[0023] (1);

[0024] Indicates the difference between the knowledge points contained in the learning resources and the knowledge points expected by students;

[0025] (2);

[0026] Indicates the difference between the difficulty of learning resources and the ability level of students;

[0027] (3);

[0028] To assess whether the time required to complete the learning resources exceeds the upper limit of the acceptable learning time for students;

[0029] (4);

[0030] in ;

[0031] To evaluate the coverage of knowledge points in learning resources;

[0032] (5);

[0033] To evaluate the balance of knowledge distribution in learning resources;

[0034] (6);

[0035] (6) is the multi-dimensional matching fitness function, where , , , and is the weight coefficient of the secondary fitness function, and the matching threshold is set ; The smaller the value, the higher the quality of the recommended learning resources, and the more suitable they are for students. The higher the value, the less suitable the learning resource is for recommending to students;

[0036] The student population is represented by S = [ s 1 , … , s n , … , s N ] , Representative Individual students, n ∈ [ 1 , … , N ] , is the number of individual students, For the The ability level of each individual student; is a binary value, if If it is 1, it means Students are expected to master Knowledge points, m ∈ [ 1 , … , M ] , Indicates the number of knowledge points, If it is 0, it means that the student does not consider mastering this knowledge point. Indicates The upper limit of the study time that a student can accept; Learning resources by express, l ∈ [ 1 , … , L ] , is the number of resources, B l = [ b l 、 1 , … , b l 、 m , … , b l 、 M ] ,like If it is 1, it means The resource contains knowledge points, 0 means that the knowledge point is not included. The difficulty of learning resources is express, Indicates The learning difficulty of each resource, Represents learning The time used by a resource; is a decision sequence, where NET n = [ Net n 、 1 , … , Net n 、 l , … , Net n 、 L ] , is a decision-making element. If it is 1, it means Resources are recommended to A student, if A value of 0 means that the student has not been recommended.

[0037] Furthermore, a non-mandatory decision element updating method is used to determine the optimal learning resources, including:

[0038] Map the element values ​​in the continuous position space to the binary discrete space. Student-related Learning Resource Decision Sequence The corresponding pigeon position vector The mapping is generated, where X k 、 n j 1 = [ x k 、 n 、 1 j 1 , … , x k 、 n 、 l j 1 , … , x k 、 n 、 L j 1 ] , Corresponding to The first Individuals, k ∈ [ 1 , … , K ] , combined with After mapping the probability value, the learning resource decision sequence is expressed as follows:

[0039] (7);

[0040] (7) In the formula, is a random number in the range [0,1];

[0041] Combined with the decision sequence results in formula (7), the fitness function value of the learning resource corresponding to the decision sequence is calculated according to the matching fitness function formula (6);

[0042] Sort the fitness function values ​​of each learning resource from small to large. The fitness function value corresponding to the optimal learning resource of each student must be the smallest, and the individual position corresponding to its decision sequence is the optimal individual in the pigeon flock, expressed as .

[0043] Furthermore, a compass guidance strategy based on the optimal solution neighborhood information is implemented to update the population individuals, including:

[0044] The speed update formula of population individuals is as follows:

[0045] (8);

[0046] Among them, During the iteration, and are all random numbers in the range [0,1]. V k 、 n j 1 = [ v k 、 n 、 1 j 1 , … , v k 、 n 、 l j 1 , … , v k 、 n 、 L j 1 ] represents the moving speed of the individual, is the moving speed of the updated individual; the historical optimal individual in the iteration process of various groups is calculated based on the matching fitness function , X n HOpt = [ x n 、 1 HOpt , … , x n 、 l HOpt , … , x n 、 L HOpt ] , randomly selected with Individuals with higher comprehensive similarity are ;

[0047] Furthermore, the formula for calculating the comprehensive similarity between individuals in the population and the best individual in history is as follows:

[0048] { Hybrid _ sim ( X n 、 k j 1 ) = ∑ l = 1 L I [ 0 , 0 . 05 ] | x n 、 l HOpt − x n 、 k 、 l j 1 | 2 L · ( ∑ l = 1 L x n 、 l HOpt − X ¯ n HOpt ) · ( ∑ l = 1 L x n 、 k 、 l j 1 − X ¯ n 、 k j 1 ) ( ∑ l = 1 L x n 、 l HOpt − X ¯ n HOpt ) 2 · ( ∑ l = 1 L x n 、 k 、 l j 1 − X ¯ n 、 k j 1 ) 2 1 ≤ n ≤ N ; 1 ≤ k ≤ K (9);

[0049] in, represents the indicator function, is the vector mean;

[0050] According to formula (9), the comprehensive similarity between the individuals in the population and the optimal individual is calculated, and the similarity values ​​are sorted from small to large. The first H% of individual positions are selected as candidate solutions, and a solution is randomly selected from them as ;

[0051] Combined with formula (8), the position update formula of the population individuals is as follows:

[0052] (10).

[0053] Furthermore, updating the positions of individual population members according to the landmark guidance includes:

[0054] According to the fitness function value sorting, half of the disadvantaged individuals in the population are eliminated, and the retained dominant individuals are expressed as , is the number of individuals in the population, During the iteration, t ∈ [ 1 , … , ⌊ 0 . 5 j 2 · K ⌋ ] , the pigeon group update formula is as follows:

[0055] (11);

[0056] (11) In the formula, In order to preserve the population of inferior individuals, is a random number in the range [0,1]. To round down, is the inverse value of the result of formula (6);

[0057] Update statistics, Do deal with;

[0058] The update discovery probability includes:

[0059] (12);

[0060] (12) In the formula, , are the maximum and minimum values ​​of the preset discovery probability respectively;

[0061] The determining whether to disturb the optimal individual comprises:

[0062] Random Generation , P rand j 2 ∈ [ 0 , 1 ] ,judge Is it greater than the probability of discovery? , if so, disturb the optimal individual in the population;

[0063] The dynamic step size local search mutation strategy based on chaotic mapping perturbs the optimal individual, including:

[0064] Determine the movement range of population individuals as follows:

[0065] (13);

[0066] In the In the population, The best individual of the contemporary era The corresponding position vector elements, For all individual positions The upper and lower extreme values ​​of the elements;

[0067] The position update formula of the optimal individual is as follows:

[0068] (14);

[0069] (14) In the formula, is the adaptive adjustment coefficient, and its expression is:

[0070] (15);

[0071] (14) In the formula, is the chaotic mapping result, and the update formula is:

[0072] (16);

[0073] (16) In the formula, is the mapping parameter.

[0074] Furthermore, the determination of whether to retain the disturbed individual includes:

[0075] According to formula (6) and formula (7), the fitness function value of the optimal individual after disturbance by the dynamic step size local search mutation strategy is calculated. If the fitness function value of the individual after disturbance is less than the value before disturbance, the position after disturbance is retained, otherwise the original position remains unchanged, that is, the position update formula of the optimal individual is equivalent to the following:

[0076] (17);

[0077] (17) In the formula, CHAOS n j 2 + 1 = [ chaos n 、 1 j 2 + 1 , … , chaos n 、 l j 2 + 1 , … , chaos n 、 L j 2 + 1 ] , is the individual position after disturbance;

[0078] The determining whether each matching function value is lower than a threshold value includes:

[0079] Combining equations (6) and (7), calculate the multidimensional matching fitness function values ​​corresponding to all individuals in each population, and determine whether the minimum function value of each population is lower than its respective matching threshold. .

[0080] In a second aspect, the present invention provides a personalized learning resource recommendation system based on an improved pigeon flock algorithm, comprising:

[0081] Acquisition module: used to obtain learners' personalized characteristics and learning needs;

[0082] Output module: used to input the personalized characteristics and learning needs into a pre-trained learning resource recommendation model, the learning resource recommendation model constructs a multi-dimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, solves the multi-dimensional matching fitness function through the improved pigeon flock algorithm and outputs the optimal learning resource sequence;

[0083] The improved pigeon flock algorithm comprises:

[0084] Randomly initialize the position and speed of the pigeon flock;

[0085] Iterate the first iteration step until the number of iterations of the first iteration step reaches a preset first maximum number of iterations, then execute the second iteration step, and output the optimal learning resource sequence by the second iteration step;

[0086] The first iteration step comprises:

[0087] Adopt non-mandatory decision element updating method to determine the optimal learning resources;

[0088] A compass-guided strategy based on the neighborhood information of the optimal solution is implemented to update the population individuals.

[0089] In a third aspect, the present invention provides a computer device, including a processor and a storage medium;

[0090] The storage medium is used to store instructions;

[0091] The processor is used to operate according to the instructions to execute the steps of the method according to the first aspect.

[0092] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0093] Compared with the prior art, the present invention has the following beneficial effects:

[0094] The personalized learning resource recommendation method based on the improved pigeon flock algorithm provided by the present invention can generate a learning resource sequence that meets the learner's ability level and interest preference according to the learner's personalized characteristics and learning needs, providing learners with a richer, more diverse and personalized learning experience. At the same time, the development of this study will also open up a new path for the theoretical development and application expansion of the pigeon flock algorithm, and promote its wide application and in-depth research in the field of smart education.

[0095] The personalized learning resource recommendation method based on the improved pigeon flock algorithm provided by the present invention innovatively defines the matching degree between learner characteristics and learning resource characteristics, learning time cost, and knowledge point coverage and distribution balance in the constructed learning resource recommendation model. In the compass operator stage of the classic pigeon flock algorithm, in order to avoid excessive use of the optimal individual, which leads to the algorithm's search behavior being too greedy and thus causing the problem of premature maturity, the present invention chooses to replace the optimal individual of the population with an individual in the neighborhood, specifically based on the concept of mutual information, randomly selects suboptimal individuals from around the optimal individual of the population to lead the pigeon flock to conduct a global search, and gives full play to the guiding role of different neighborhood information on the algorithm search. In the classic landmark operator stage, in order to enhance the anti-hysteresis of the algorithm, the present invention introduces an operation based on the probability of discovery to eliminate poor solutions to improve the diversity of the population, specifically introducing a dynamic step-size local search mutation strategy based on chaotic mapping to improve the algorithm's local extreme value escape ability. In order to balance the global exploration and local development capabilities in the population search process, the present invention designs an adaptive probability formula, and the probability value changes dynamically with the increase in the number of algorithm iterations to improve the convergence performance of the algorithm in the later stage. Compared with the existing learning resource recommendation method based on swarm intelligence optimization algorithm, the personalized learning resource recommendation method of the present invention has higher recommendation accuracy, stability and learning path matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 It is the algorithm flow chart of the present invention;

[0097] Figure 2 It is a comparison of convergence curves of different algorithms in the experimental environment of the embodiment of the present invention;

[0098] Figure 3 It is a comparison of convergence curves of different algorithms in the environment of Experiment 2 in the embodiment of the present invention;

[0099] Figure 4 It is a comparison of convergence curves of different algorithms in the third experimental environment of the embodiment of the present invention;

[0100] Figure 5 It is a comparison of convergence curves of different algorithms in the fourth experimental environment of the embodiment of the present invention;

[0101] Figure 6 It is a comparison of convergence curves of different algorithms in the fifth environment of the experiment in the embodiment of the present invention;

[0102] Figure 7 It is a flow chart of the personalized learning resource recommendation method based on the improved pigeon flock algorithm provided by the present invention. DETAILED DESCRIPTION

[0103] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0104] The term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0105] Embodiment 1:

[0106] Figure 7 This is a flowchart of a personalized learning resource recommendation method based on an improved pigeon flock algorithm in the first embodiment of the present invention. The personalized learning resource recommendation method based on the improved pigeon flock algorithm provided in this embodiment can be applied to a terminal and can be executed by a personalized learning resource recommendation system based on the improved pigeon flock algorithm. The system can be implemented by software and / or hardware. The system can be integrated in a terminal, for example: any smart phone, tablet computer or computer device with communication function. Figure 7 , the method of this implementation includes:

[0107] Obtain learners' personalized characteristics and learning needs;

[0108] The personalized characteristics and learning needs are input into a pre-trained learning resource recommendation model, the learning resource recommendation model constructs a multi-dimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, solves the multi-dimensional matching fitness function through an improved pigeon flock algorithm and outputs an optimal learning resource sequence;

[0109] Specifically include:

[0110] Step A: Based on the students' ability level, expected learning points and upper limit of learning time, the knowledge points included in the learning resources, resource difficulty and the time required for learning resources, construct a multi-dimensional matching fitness function between students and personalized learning resources, and set the matching threshold. Randomly initialize the position and speed of the pigeon flock and set the maximum number of iterations in the compass guidance phase and the hybrid optimization phase;

[0111] Step B: Use the non-mandatory decision element update method to determine the optimal learning resources;

[0112] Step C: Implement the compass guidance strategy based on the optimal solution neighborhood information to update the population individuals;

[0113] Step D: Determine whether the compass guidance phase of the algorithm has reached the maximum number of iterations. If so, execute step E; otherwise, execute step B.

[0114] Step E: Update the individual positions of the population based on landmark guidance;

[0115] Step F: Update the discovery probability;

[0116] Step G: Determine whether to disturb the optimal individual. If so, execute step H, otherwise execute step I;

[0117] Step H: Implement a dynamic step-size local search mutation strategy based on chaotic mapping to perturb the optimal individual and determine whether to retain the perturbed individual;

[0118] Step I: Determine whether the values ​​of each matching function are lower than the threshold or whether the hybrid optimization stage of the algorithm has reached the maximum number of iterations. If any of the conditions are met, execute step J, otherwise execute step E;

[0119] Step J: Output optimal learning resources.

[0120] The present invention proposes a personalized learning resource recommendation method based on an improved pigeon flock algorithm. The pigeon flock algorithm, as an emerging heuristic optimization algorithm, demonstrates the ability to efficiently search for optimal solutions in complex nonlinear problems by virtue of its unique mechanism of simulating the foraging and homing behavior of pigeon flocks in nature. The algorithm has attracted widespread attention in the academic community for its simple and intuitive principle, strong robustness and fast convergence speed. However, like other swarm intelligence algorithms, the pigeon flock algorithm also faces the dilemma of insufficient convergence accuracy and easy to fall into local optimality. To this end, this study is committed to in-depth improvement and innovation of the pigeon flock algorithm, aiming to overcome its existing defects and enhance its application efficiency in the field of personalized learning resource recommendation in smart education;

[0121] Step Aa: The student population is represented as S = [ s 1 , … , s n , … , s N ] , Representative Individual students, n ∈ [ 1 , … , N ] , is the number of individual students, For the The ability level of individual students. is a binary value, if If it is 1, it means Students are expected to master Knowledge points, m ∈ [ 1 , … , M ] , Indicates the number of knowledge points, If it is 0, it means that the student does not consider mastering this knowledge point. Indicates The upper limit of the study time that a student can accept. Learning resources by express, l ∈ [ 1 , … , L ] , is the number of resources, B l = [ b l 、 1 , … , b l 、 m , … , b l 、 M ] ,like If it is 1, it means The resource contains knowledge points, 0 means that the knowledge point is not included. The difficulty of learning resources is express, Indicates The learning difficulty of each resource, Represents learning The time used by a resource. is a decision sequence, where NET n = [ Net n 、 1 , … , Net n 、 l , … , Net n 、 L ] , is a decision-making element. If it is 1, it means Resources are recommended to A student, if If it is 0, it means that it is not recommended to the student. In order to make personalized learning resources match learner characteristics and improve the quality of personalized learning resource generation services, five secondary fitness functions are established as follows:

[0122] (1);

[0123] Indicates the difference between the knowledge points contained in the learning resources and the knowledge points expected by students.

[0124] (2);

[0125] Represents the difference between the difficulty of learning resources and the ability level of students.

[0126] (3);

[0127] To assess whether the time required to complete the learning resources exceeds the upper limit of the acceptable learning time for students.

[0128] (4);

[0129] in ;

[0130] To evaluate the coverage of knowledge points in learning resources.

[0131] (5);

[0132] To evaluate the balanced distribution of knowledge points in learning resources.

[0133] (6);

[0134] (6) is the multi-dimensional matching fitness function, where , , , and is the weight coefficient of the secondary fitness function, and the matching threshold is set .

[0135] Step Ab: Random Generation between [0,1] dimensional vector as the The initial position vector of the pigeon group, Indicates the number of individuals in the population, and randomly generates the same number of values ​​between [0,1] as the initial speed of the corresponding individuals. Set the maximum number of iterations in the compass guidance phase and the maximum number of iterations in the hybrid optimization phase , It is the statistical value of the number of iterations in the compass guidance phase. It is the statistical value of the number of iterations in the hybrid optimization phase.

[0136] Step Ba: The specific operation of the non-mandatory decision element update method is as follows: Map the element value in the continuous position space to the binary discrete space. Student-related Learning Resource Decision Sequence The corresponding pigeon position vector The mapping is generated, where X k 、 n j 1 = [ x k 、 n 、 1 j 1 , … , x k 、 n 、 l j 1 , … , x k 、 n 、 L j 1 ] , Corresponding to The first Individuals, k ∈ [ 1 , … , K ] , combined with After mapping the probability value, the learning resource decision sequence is expressed as follows:

[0137] (7);

[0138] (7) In the formula, is a random number in the range [0,1].

[0139] Combined with the decision sequence results in formula (7), the fitness function value of the learning resource corresponding to the decision sequence is calculated according to the matching fitness function formula (6).

[0140] Step Bb: The specific operation of determining the optimal learning resource is: sort the fitness function values ​​of each learning resource from small to large, The fitness function value corresponding to the optimal learning resource of each student must be the smallest, and the individual position corresponding to its decision sequence is the optimal individual in the pigeon flock, expressed as .

[0141] Step Bc: Update the statistics. Do deal with.

[0142] Step Ca: The specific operation of updating the population individuals based on the compass guidance strategy of the optimal solution neighborhood information is as follows:

[0143] The speed update formula of population individuals is as follows:

[0144] (8);

[0145] Among them, During the iteration, and are all random numbers in the range [0,1]. V k 、 n j 1 = [ v k 、 n 、 1 j 1 , … , v k 、 n 、 l j 1 , … , v k 、 n 、 L j 1 ] represents the moving speed of the individual, is the moving speed of the updated individual. The historical optimal individual in the iteration process of various groups is calculated based on the matching fitness function. , X n HOpt = [ x n 、 1 HOpt , … , x n 、 l HOpt , … , x n 、 L HOpt ] , randomly selected with Individuals with higher comprehensive similarity are .

[0146] Step Cb: The formula for calculating the comprehensive similarity between individuals in the population and the historical best individual is as follows:

[0147] { Hybrid _ sim ( X n 、 k j 1 ) = ∑ l = 1 L I [ 0 , 0 . 05 ] | x n 、 l HOpt − x n 、 k 、 l j 1 | 2 L · ( ∑ l = 1 L x n 、 l HOpt − X ¯ n HOpt ) · ( ∑ l = 1 L x n 、 k 、 l j 1 − X ¯ n 、 k j 1 ) ( ∑ l = 1 L x n 、 l HOpt − X ¯ n HOpt ) 2 · ( ∑ l = 1 L x n 、 k 、 l j 1 − X ¯ n 、 k j 1 ) 2 1 ≤ n ≤ N ; 1 ≤ k ≤ K (9);

[0148] in, represents the indicator function, is the vector mean.

[0149] According to formula (9), the comprehensive similarity between the individuals in the population and the optimal individual is calculated, and the similarity values ​​are sorted from small to large. The first H% of individual positions are selected as candidate solutions, and a solution is randomly selected from them as .

[0150] Step Cc: Combined with formula (8), the position update formula of the population individuals is as follows:

[0151] (10);

[0152] Step D: Determine whether the compass guidance phase of the algorithm has reached the maximum number of iterations. Is it equal to the maximum number of iterations? .

[0153] Step Ea: The specific operation of updating the position of individuals in the population based on landmark guidance is as follows: sorting by fitness function value, removing half of the inferior individuals in the population, and the retained superior individuals are represented as , is the number of individuals in the population, During the iteration, t ∈ [ 1 , … , ⌊ 0 . 5 j 2 · K ⌋ ] , the pigeon group update formula is as follows:

[0154] (11);

[0155] (11) In the formula, In order to preserve the population of inferior individuals, is a random number in the range [0,1]. To round down, is the inverse of the result of formula (6).

[0156] Step Eb: Update the statistical value. Do deal with.

[0157] Step F: The specific operations for updating the discovery probability are as follows:

[0158] (12);

[0159] (12) In the formula, , are the maximum and minimum values ​​of the preset discovery probability respectively.

[0160] Step G: The specific operation of determining whether to disturb the optimal individual is as follows:

[0161] Random Generation , P rand j 2 ∈ [ 0 , 1 ] ,judge Is it greater than the probability of discovery? If so, the best individual in the population is disturbed.

[0162] Step Ha: The specific operation of the dynamic step size local search mutation strategy based on chaotic mapping to perturb the optimal individual is as follows:

[0163] Determine the movement range of population individuals as follows:

[0164] (13);

[0165] In the In the population, The best individual of the contemporary era The corresponding position vector elements, For all individual positions The upper and lower extreme values ​​of an element.

[0166] The position update formula of the optimal individual is as follows:

[0167] (14);

[0168] (14) In the formula, is the adaptive adjustment coefficient, and its expression is:

[0169] (15);

[0170] (14) In the formula, is the chaotic mapping result, and the update formula is:

[0171] (16);

[0172] (16) In the formula, is the mapping parameter.

[0173] Step Hb: The detailed steps of judging whether to retain the individual after the disturbance are as follows: According to formula (6) and formula (7), the fitness function value of the optimal individual after the disturbance by the dynamic step size local search mutation strategy is calculated. If the fitness function value of the individual after the disturbance is less than the value before the disturbance, the position after the disturbance is retained, otherwise the original position remains unchanged, that is, the position update formula of the optimal individual is equivalent to the following:

[0174] (17);

[0175] (17) In the formula, CHAOS n j 2 + 1 = [ chaos n 、 1 j 2 + 1 , … , chaos n 、 l j 2 + 1 , … , chaos n 、 L j 2 + 1 ] , is the individual position after disturbance.

[0176] Step Ia: Determine whether all matching function values ​​are lower than the threshold. The specific operation is as follows: Combine equation (6) and equation (7) to calculate the multidimensional matching fitness function values ​​corresponding to all individuals in each population, and determine whether the minimum function value of each population is lower than the respective matching threshold. .

[0177] Step Ib: Determine whether the hybrid optimization phase of the algorithm has reached the maximum number of iterations. The specific operation is as follows: Determine Is it equal to the maximum number of iterations? .

[0178] Step J: The specific operation of outputting the optimal learning resource is as follows: According to formula (6), the multi-dimensional matching fitness function value corresponding to the learning resource to be output needs to be the smallest.

[0179] S10: Establish multi-dimensional matching fitness function and set matching threshold

[0180] The student population is represented by S = [ s 1 , … , s n , … , s N ] , Representative Individual students, n ∈ [ 1 , … , N ] , is the number of individual students, For the The ability level of individual students. is a binary value, if If it is 1, it means Students are expected to master Knowledge points, m ∈ [ 1 , … , M ] , Indicates the number of knowledge points, If it is 0, it means that the student does not consider mastering this knowledge point. Indicates The upper limit of the study time that a student can accept. Learning resources by express, l ∈ [ 1 , … , L ] , is the number of resources, B l = [ b l 、 1 , … , b l 、 m , … , b l 、 M ] ,like If it is 1, it means The resource contains knowledge points, 0 means that the knowledge point is not included. The difficulty of learning resources is express, Indicates The learning difficulty of each resource, Represents learning The time used by a resource. is a decision sequence, where NET n = [ Net n 、 1 , … , Net n 、 l , … , Net n 、 L ] , is a decision-making element. If it is 1, it means Resources are recommended to A student, if If it is 0, it means that it is not recommended to the student. In order to make personalized learning resources match learner characteristics and improve the quality of personalized learning resource generation services, five secondary fitness functions are established as follows:

[0181] (1);

[0182] Indicates the difference between the knowledge points contained in the learning resources and the knowledge points expected by students.

[0183] (2);

[0184] Represents the difference between the difficulty of learning resources and the ability level of students.

[0185] (3);

[0186] To assess whether the time required to complete the learning resources exceeds the upper limit of the acceptable learning time for students.

[0187] (4);

[0188] in

[0189] To evaluate the coverage of knowledge points in learning resources.

[0190] (5);

[0191] To evaluate the balanced distribution of knowledge points in learning resources.

[0192] (6);

[0193] (6) is the multi-dimensional matching fitness function, where , , , and is the weight coefficient of the secondary fitness function, and the matching threshold is set .

[0194] S11: Randomly initialize the position and speed of the pigeon flock, set the maximum number of iterations in the compass guidance phase and the hybrid optimization phase

[0195] Random Generation between [0,1] dimensional vector as the The initial position vector of the pigeon group, Indicates the number of individuals in the population, and randomly generates the same number of values ​​between [0,1] as the initial speed of the corresponding individuals. Set the maximum number of iterations in the compass guidance phase and the maximum number of iterations in the hybrid optimization phase , It is the statistical value of the number of iterations in the compass guidance phase. It is the statistical value of the number of iterations in the hybrid optimization phase.

[0196] S12: Using non-mandatory decision element updating method to determine the optimal learning resources

[0197] Map the element values ​​in the continuous position space to the binary discrete space. Student-related Learning Resource Decision Sequence The corresponding pigeon position vector The mapping is generated, where X k 、 n j 1 = [ x k 、 n 、 1 j 1 , … , x k 、 n 、 l j 1 , … , x k 、 n 、 L j 1 ] , Corresponding to The first Individuals, k ∈ [ 1 , … , K ] , combined with After mapping the probability value, the learning resource decision sequence is expressed as follows:

[0198] (7);

[0199] (7) In the formula, is a random number in the range [0,1].

[0200] Combined with the decision sequence results in formula (7), the fitness function value of the learning resource corresponding to the decision sequence is calculated according to the matching fitness function formula (6).

[0201] Sort the fitness function values ​​of each learning resource from small to large. The fitness function value corresponding to the optimal learning resource of each student must be the smallest, and the individual position corresponding to its decision sequence is the optimal individual in the pigeon flock, expressed as .

[0202] Update statistics, Do deal with.

[0203] S13: Implementing a compass-guided strategy based on the optimal solution neighborhood information to update population individuals

[0204] The speed update formula of population individuals is as follows:

[0205] (8);

[0206] Among them, During the iteration, and are all random numbers in the range [0,1]. V k 、 n j 1 = [ v k 、 n 、 1 j 1 , … , v k 、 n 、 l j 1 , … , v k 、 n 、 L j 1 ] represents the moving speed of the individual, is the moving speed of the updated individual. The historical optimal individual in the iteration process of various groups is calculated based on the matching fitness function. , X n HOpt = [ x n 、 1 HOpt , … , x n 、 l HOpt , … , x n 、 L HOpt ] , randomly selected with Individuals with higher comprehensive similarity are .

[0207] The formula for calculating the comprehensive similarity between individuals in the population and the best individual in history is as follows:

[0208] { Hybrid _ sim ( X n 、 k j 1 ) = ∑ l = 1 L I [ 0 , 0 . 05 ] | x n 、 l HOpt − x n 、 k 、 l j 1 | 2 L · ( ∑ l = 1 L x n 、 l HOpt − X ¯ n HOpt ) · ( ∑ l = 1 L x n 、 k 、 l j 1 − X ¯ n 、 k j 1 ) ( ∑ l = 1 L x n 、 l HOpt − X ¯ n HOpt ) 2 · ( ∑ l = 1 L x n 、 k 、 l j 1 − X ¯ n 、 k j 1 ) 2 1 ≤ n ≤ N ; 1 ≤ k ≤ K (9);

[0209] in, represents the indicator function, is the vector mean.

[0210] According to formula (9), the comprehensive similarity between the individuals in the population and the optimal individual is calculated, and the similarity values ​​are sorted from small to large. The first H% of individual positions are selected as candidate solutions, and a solution is randomly selected from them as .

[0211] Combined with formula (8), the position update formula of the population individuals is as follows:

[0212] (10);

[0213] S14: Determine whether the compass guidance phase of the algorithm has reached the maximum number of iterations

[0214] judge Is it equal to the maximum number of iterations? .

[0215] S15: Update the position of individual populations based on landmark guidance

[0216] According to the fitness function value sorting, half of the disadvantaged individuals in the population are eliminated, and the retained dominant individuals are expressed as , is the number of individuals in the population, During the iteration, t ∈ [ 1 , … , ⌊ 0 . 5 j 2 · K ⌋ ] , the pigeon group update formula is as follows:

[0217] (11);

[0218] (11) In the formula, In order to preserve the population of inferior individuals, is a random number in the range [0,1]. To round down, is the inverse of the result of formula (6).

[0219] Further, update the statistical values. Do deal with.

[0220] S16: Update discovery probability

[0221] (12);

[0222] (12) In the formula, , are the maximum and minimum values ​​of the preset discovery probability respectively.

[0223] S17: Determine whether to disturb the optimal individual

[0224] Random Generation , P rand j 2 ∈ [ 0 , 1 ] ,judge Is it greater than the probability of discovery? If so, execute S18 to perturb the best individual in the population.

[0225] S18: Dynamic step-size local search mutation strategy based on chaotic mapping to perturb the optimal individual

[0226] Determine the movement range of population individuals as follows:

[0227] (13);

[0228] In the In the population, The best individual of the contemporary era The corresponding position vector elements, For all individual positions The upper and lower extreme values ​​of an element.

[0229] The position update formula of the optimal individual is as follows:

[0230] (14);

[0231] (14) In the formula, is the adaptive adjustment coefficient, and its expression is:

[0232] (15);

[0233] (14) In the formula, is the chaotic mapping result, and the update formula is:

[0234] (16);

[0235] (16) In the formula, is the mapping parameter.

[0236] S19: Determine whether the fitness function value of the individual becomes smaller after the disturbance

[0237] If the fitness function value corresponding to the optimal individual after disturbance by the dynamic step size local search mutation strategy based on chaotic mapping is less than the fitness value before disturbance, S20 is executed, otherwise S21 is executed.

[0238] S20: retain the disturbed individuals

[0239] The optimal individual retains the disturbed position, which is equivalent to updating the position of the optimal individual according to formula (17).

[0240] (17);

[0241] S21: The optimal individual position remains unchanged

[0242] This is equivalent to equation (18), where the optimal individual maintains its position before being disturbed.

[0243] (18);

[0244] S22: Determine whether the values ​​of each multi-dimensional matching fitness function are all lower than the threshold or whether the hybrid optimization stage of the algorithm has reached the maximum number of iterations

[0245] Combining equations (6) and (7), calculate the multidimensional matching fitness function values ​​corresponding to all individuals in each population, and determine whether the minimum function value of each population is lower than its respective matching threshold. ,judge Is it equal to the maximum number of iterations? .

[0246] S23: Output the best learning resources

[0247] According to formula (6), the multi-dimensional matching fitness function value corresponding to the output learning resource must be minimized.

[0248] Experiment: Select references ([1] Chu, Chih Ping, YCChang, and C.C. Tsai. "PC2PSO: personalized e-course composition based on Particle Swarm Optimization." Applied Intelligence 34.1 (2011): 141-154. [2] Dheeban SG, Deepak V, DhamodharanL, et al. Improved personalized e-course composition approach using modified particle swarm optimization with inertia-coefficient[J]. InternationalJournal of Computer Applications, 2010, 1(6): 102-107.[3]Mirjalili S, LewisA.S-shaped versus V-shaped transfer functions for binary particle swarmoptimization[J].Swarm and Evolutionary Computation, 2013, 9: 1-14. [4] Li Haojun, Liu Zhongfeng, Li Sai, et al. Personalized network learning resource recommendation method based on improved binary particle swarm algorithm [J]. Systems Science and Mathematics, 2017, 37(08): 1770-1779.) The examples in the paper are compared with the results of the present invention.

[0249] In order to verify the performance of the modified pigeon-inspired optimization (MPIO) algorithm in the personalized learning resource recommendation problem, five groups of experiments were implemented. The unified parameter settings are shown in Table 1, and the specific learner characteristics and learning resource characteristics are shown in Table 2. The relevant parameter settings in the competitive algorithm are consistent with those in the literature [4].

[0250] Table 1 Parameter settings of the improved pigeon flock algorithm in the present invention

[0251] Table 2 Learner characteristics and learning resource characteristics

[0252] After 50 independent experiments, the mean and variance data of each algorithm after convergence are shown in Table 3. The convergence curve corresponding to the optimal learning resource is shown in Figure 2-6 As shown in Table 3, under different learner characteristics and learning resource characteristics, the MPIO algorithm in the present invention has the best convergence accuracy and stability. From the convergence curve in the figure, it can be seen that when the MPIO algorithm solves the problem of personalized learning resource recommendation, the multi-dimensional matching degree corresponding to the obtained learning resources is better than that of the competing algorithm, and has a faster convergence speed.

[0253] Table 3 Convergence mean and variance

[0254] Embodiment 2:

[0255] A personalized learning resource recommendation system based on an improved pigeon flock algorithm, comprising: Acquisition module: used to obtain learners' personalized characteristics and learning needs; Output module: used to input the personalized characteristics and learning needs into a pre-trained learning resource recommendation model, the learning resource recommendation model constructs a multi-dimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, solves the multi-dimensional matching fitness function through the improved pigeon flock algorithm and outputs the optimal learning resource sequence; The improved pigeon flock algorithm comprises: Randomly initialize the position and speed of the pigeon flock and set the maximum number of iterations in the compass guidance phase and the hybrid optimization phase; Adopt non-mandatory decision element updating method to determine the optimal learning resources; Implement a compass-guided strategy based on the optimal solution neighborhood information to update population individuals; Determine whether the compass guidance phase of the algorithm has reached the maximum number of iterations. If so, execute the iteration step, otherwise use the non-mandatory decision element update method to determine the optimal learning resource.

[0256] The personalized learning resource recommendation system based on the improved pigeon flock algorithm provided in the embodiment of the present invention can execute the personalized learning resource recommendation method based on the improved pigeon flock algorithm provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0257] Embodiment three: An embodiment of the present invention further provides a computer device, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to perform the steps of the following method: Obtain learners' personalized characteristics and learning needs; The personalized characteristics and learning needs are input into a pre-trained learning resource recommendation model. The learning resource recommendation model constructs a multidimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, and solves the multidimensional matching fitness function through the improved pigeon flock algorithm and outputs the optimal learning resource sequence.

[0258] For the details of each step in this embodiment, please refer to the first embodiment, which will not be described in detail here. Since this embodiment and the first embodiment adopt the same technical concept, it also has the technical effects described in the first embodiment.

[0259] Embodiment 4: The embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the following method are implemented: Obtain learners' personalized characteristics and learning needs; The personalized characteristics and learning needs are input into a pre-trained learning resource recommendation model. The learning resource recommendation model constructs a multidimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, and solves the multidimensional matching fitness function through the improved pigeon flock algorithm and outputs the optimal learning resource sequence.

[0260] For the details of each step in this embodiment, please refer to the first embodiment, which will not be described in detail here. Since this embodiment and the first embodiment adopt the same technical concept, it also has the technical effects described in the first embodiment.

[0261] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0262] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or multiple boxes.

[0263] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction system, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0264] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0265] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A personalized learning resource recommendation method based on an improved pigeon flock algorithm, characterized in that: include: Obtain learners' personalized characteristics and learning needs; The personalized characteristics and learning needs are input into a pre-trained learning resource recommendation model, the learning resource recommendation model constructs a multi-dimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, solves the multi-dimensional matching fitness function through an improved pigeon flock algorithm and outputs an optimal learning resource sequence; The improved pigeon flock algorithm comprises: Randomly initialize the position and speed of the pigeon flock; Iterate the first iteration step until the number of iterations of the first iteration step reaches a preset first maximum number of iterations, then execute the second iteration step, and output the optimal learning resource sequence by the second iteration step; The first iteration step comprises: Adopt non-mandatory decision element updating method to determine the optimal learning resources; A compass-guided strategy based on the neighborhood information of the optimal solution is implemented to update the population individuals.

2. The personalized learning resource recommendation method based on the improved pigeon flock algorithm according to claim 1 is characterized in that: The second iteration step comprises: Update the individual positions of the population based on landmark guidance; Update discovery probability; Determine whether to disturb the optimal individual. If so, implement the dynamic step size local search mutation strategy based on chaotic mapping to disturb the optimal individual. Determine whether to retain the disturbed individual. Otherwise, determine whether the values ​​of each matching function are lower than the threshold or whether the second iteration step reaches the preset second maximum number of iterations. If any of the conditions are met, output the optimal learning resource sequence, otherwise re-execute the second iteration step.

3. The personalized learning resource recommendation method based on the improved pigeon flock algorithm according to claim 2 is characterized in that: The multidimensional matching fitness function includes five secondary fitness functions, among which, (1); is the first secondary fitness function, which indicates the difference between the knowledge points contained in the learning resources and the knowledge points expected by the students; (2); is the second secondary fitness function, which represents the difference between the difficulty of learning resources and the ability level of students; (3); is the third-level fitness function, which is used to evaluate whether the time required to complete the learning resources exceeds the upper limit of the acceptable learning time for students; (4); in ; is the fourth-order fitness function, which is used to evaluate the coverage of knowledge points in learning resources; (5); is the fifth-order fitness function, which is used to evaluate the distribution balance of knowledge points in learning resources; (6); (6) is the multi-dimensional matching fitness function, where , , , and is the weight coefficient of the secondary fitness function, and the matching threshold is set ; The smaller the value, the higher the quality of the recommended learning resources, and the more suitable they are for students. The higher the value, the less suitable the learning resource is for recommending to students; The student population is represented by , Representative Individual students, , is the number of individual students, For the The ability level of each individual student; is a binary value, if If it is 1, it means Students are expected to master Knowledge points, , Indicates the number of knowledge points, If it is 0, it means that the student does not consider mastering this knowledge point. Indicates The upper limit of the study time that a student can accept; Learning resources by express, , is the number of resources, ,like If it is 1, it means The resource contains knowledge points, 0 means that the knowledge point is not included. The difficulty of learning resources is express, Indicates The learning difficulty of each resource, Represents learning The time used by a resource; is a decision sequence, where , is a decision-making element. If it is 1, it means Resources are recommended to A student, if A value of 0 means that the student has not been recommended.

4. The personalized learning resource recommendation method based on the improved pigeon flock algorithm according to claim 3 is characterized in that: The optimal learning resources are determined using a non-mandatory decision element update method, including: Map the element values ​​in the continuous position space to the binary discrete space. Student-related Learning Resource Decision Sequence The corresponding pigeon position vector The mapping is generated, where , Corresponding to The first Individuals, , combined with After mapping the probability value, the learning resource decision sequence is expressed as follows: (7); (7) In the formula, is a random number in the range [0,1]; Combined with the decision sequence results in formula (7), the fitness function value of the learning resource corresponding to the decision sequence is calculated according to the matching fitness function formula (6); Sort the fitness function values ​​of each learning resource from small to large. The fitness function value corresponding to the optimal learning resource of each student must be the smallest, and the individual position corresponding to its decision sequence is the optimal individual in the pigeon flock, which is expressed as .

5. The personalized learning resource recommendation method based on the improved pigeon flock algorithm according to claim 4 is characterized in that: Implement a compass-guided strategy based on the optimal solution neighborhood information to update population individuals, including: The speed update formula of population individuals is as follows: (8); Among them, During the iteration, and are all random numbers in the range [0,1]. represents the moving speed of the individual, is the moving speed of the updated individual; the historical optimal individual in the iteration process of various groups is calculated based on the matching fitness function , , randomly selected with Individuals with higher comprehensive similarity are ; The formula for calculating the comprehensive similarity between individuals in the population and the best individual in history is as follows: (9); in, represents the indicator function, is the vector mean; According to formula (9), the comprehensive similarity between the individuals in the population and the optimal individual is calculated, and the similarity values ​​are sorted from small to large. The first H% of individual positions are selected as candidate solutions, and a solution is randomly selected from them as ; Combined with formula (8), the position update formula of the population individuals is as follows: (10); In the formula, is the current iteration number of the first iteration step.

6. The personalized learning resource recommendation method based on the improved pigeon flock algorithm according to claim 5 is characterized in that: The updating of the positions of the individuals in the population according to the landmark guidance includes: According to the fitness function value sorting, half of the disadvantaged individuals in the population are eliminated, and the retained dominant individuals are expressed as , is the number of individuals in the population, During the iteration, , the pigeon group update formula is as follows: (11); (11) In the formula, In order to preserve the population of inferior individuals, is a random number in the range [0,1]. To round down, is the inverse value of the result of formula (6); Update statistics, Do deal with; The update discovery probability includes: (12); (12) In the formula, , are the maximum and minimum values ​​of the preset discovery probability respectively; The determining whether to disturb the optimal individual comprises: Random Generation , ,judge Is it greater than the probability of discovery? , if so, disturb the optimal individual in the population; The dynamic step size local search mutation strategy based on chaotic mapping perturbs the optimal individual, including: Determine the movement range of population individuals as follows: (13); In the In the population, The best individual of the contemporary era The corresponding position vector elements, For all individual positions The upper and lower extreme values ​​of the elements; The position update formula of the optimal individual is as follows: (14); (14) In the formula, is the adaptive adjustment coefficient, and its expression is: (15); (14) In the formula, is the chaotic mapping result, and the update formula is: (16); (16) In the formula, is the mapping parameter.

7. The personalized learning resource recommendation method based on the improved pigeon flock algorithm according to claim 6 is characterized in that: The determining whether to retain the disturbed individual comprises: According to formula (6) and formula (7), the fitness function value of the optimal individual after disturbance by the dynamic step size local search mutation strategy is calculated. If the fitness function value of the individual after disturbance is less than the value before disturbance, the position after disturbance is retained, otherwise the original position remains unchanged, that is, the position update formula of the optimal individual is equivalent to the following: (17); (17) In the formula, , is the individual position after disturbance; The determining whether each matching function value is lower than a threshold value includes: Combining equations (6) and (7), calculate the multidimensional matching fitness function values ​​corresponding to all individuals in each population, and determine whether the minimum function value of each population is lower than its respective matching threshold. .

8. A personalized learning resource recommendation system based on an improved pigeon flock algorithm, characterized in that: include: Acquisition module: used to obtain learners' personalized characteristics and learning needs; Output module: used to input the personalized characteristics and learning needs into a pre-trained learning resource recommendation model, the learning resource recommendation model constructs a multi-dimensional matching fitness function between the learner and the personalized learning resources according to the learner's personalized characteristics and learning needs, solves the multi-dimensional matching fitness function through the improved pigeon flock algorithm and outputs the optimal learning resource sequence; The improved pigeon flock algorithm comprises: Randomly initialize the position and speed of the pigeon flock; Iterate the first iteration step until the number of iterations of the first iteration step reaches a preset first maximum number of iterations, then execute the second iteration step, and output the optimal learning resource sequence by the second iteration step; The first iteration step comprises: Adopt non-mandatory decision element updating method to determine the optimal learning resources; A compass-guided strategy based on the neighborhood information of the optimal solution is implemented to update the population individuals.

9. A computer device, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are implemented.

Citation Information

Patent Citations

  • Unmanned aerial vehicle three-dimensional path planning method based on multi-strategy improved particle swarm optimization algorithm

    CN109269502A

  • Multi-metal strip mine multi-target ore blending method based on particle swarm-pigeon swarm hybrid optimization algorithm

    CN110533246A

  • Airplane flight path planning method and device based on the pigeon-inspired optimization

    US20190035286A1

  • Methods and apparatus for application prediction through machine learning based analysis of IO patterns

    US20200372401A1

Cited By

  • Online learning resource recommendation method and system based on improved badger population algorithm

    CN120296243A

  • An online learning resource recommendation method and system based on an improved meerkat population algorithm

    CN120296243B

  • Intelligent policy resource matching method and system

    CN120407795A

  • Self-learning path generation method based on adaptive multi-objective locust optimization algorithm

    CN120448648A

  • Personalized wrong question review resource generation method and system based on knowledge graph

    CN122087195A