基于随机游走和多样性图排序的个性化服务推荐方法

针对传统服务推荐算法由于数据稀疏性而导致推荐准确性不高,以及推荐结果缺乏多样性等缺陷,提出基于随机游走和多样性图排序的个性化服务推荐方法（PRWDR）.在分析直接相似关系稀疏性的基础上提出带权重的随机游走模型,通过在用户网络上进行随机游走来挖掘更多的相似关系;基于所有相似用户预测服务的QoS值,并给出服务图模型构建方法,以过滤大量性能过低的候选服务;提出最优节点集合选取策略,利用贪婪算法得到兼具推荐准确性和功能多样性的服务推荐列表.在公开发布的数据集上进行实验,并与多个经典算法进行比较,验证了本算法的有效性.     

基于协同过滤算法的习题系统的设计与实现

本文针对现有习题系统题目搜索手段的不足,提出了构建基于协同过滤算法的习题系统。着重介绍了协同过滤算法的相关知识,详细阐述了推荐引擎的具体实现步骤。此系统在分析用户历史行为及题目相关性的基础上,可为用户提供精准和个性化的推荐结果,对提高学生的学习效率起到了积极的作用。     

大数据分析的智慧景点智能推荐模型

为了保证游客对推荐景点的满意度，以西安寺院为例，提出了大数据分析下的智慧景点智能推荐模型。采用Petri网景点游客分流模型获取超载景点以及可分流的目标景点集合，得出符合规则的目标分流游客信息集合后，通过协同过滤算法和情景上下文信息的结合，挖掘该信息集合内游客基本属性相似度、外部环境相似度以及综合相似度，依据评分值形成景点推荐集，完成智慧景点智能推荐。测试结果表明：该模型可有效完成游客负载分流，折扣累积利润和排序偏差准确率值均在0.9以上，实现游客数量最大化接纳，提升景点管理水平。     

基于加权关联规则挖掘算法的电子商务商品推荐系统研究

为了解决电子购物者和商家直接的商品快速、准确匹配问题,进行基于加权关联规则挖掘算法的电子商务商品推荐系统研究。首先指出了经典Apriori算法的缺点和不足,并提出一种新的加权模糊关联挖掘模型算法,以保证频繁项集的向下封闭性;通过对电子商务推荐系统的结构化设计、数据预处理模块设计、推荐模块设计,完成了推荐系统的工作流程测试;最后选取命中率作为不同推荐模型的评价标准,通过五折交叉试验法对实际采集数据进行了对比分析,试验结果表明关联规则集的Top-N产品命中率要明显高于兴趣推荐和畅销推荐法。     

基于Mahout与协同过滤算法的中医调理文章推荐引擎

本文研究了推荐系统的相关知识,在传统的基于用户的协同过滤算法上引入了属性相似度,增加了相似度的准确度,提高了推荐性能;利用评估推荐器计算平均绝对误差评测、调整推荐器的性能;利用Mahout开源框架,结合协同过滤算法构建了中医调理文章推荐系统。     

基于LDA主题模型的协同过滤推荐算法

传统的协同过滤推荐算法直接根据用户对物品的评分进行推荐,忽略了评论文本中隐含的重要信息,当用户对物品的评论较少时,由于数据的稀疏性会造成推荐效果的不准确和单一。本文提出了一种基于LDA主题模型的协同过滤推荐算法LDA-CF(Latent Dirichlet Allocation model-LDA-Collaborative Filtering),在传统的协同过滤算法基础上,通过LDA模型对评论文本中的主题进行分类,从各个主题层面挖掘用户的情感偏好,计算用户之间的相似度,进而向目标用户推荐商品。对京东平台牙膏的评论数据集的实验结果表明,该算法不仅可以缓解由于评分数据较少造成的稀疏性问题,推荐的精确度也有所提高。     

基于移动用户上下文相似度的协同过滤推荐算法

该文面向移动通信网络领域的个性化服务推荐问题,通过将移动用户上下文信息引入协同过滤推荐过程,提出一种基于移动用户上下文相似度的改进协同过滤推荐算法。该算法首先计算基于移动用户的上下文相似度,以构造目标用户当前上下文的相似上下文集合,然后采用上下文预过滤推荐方法对移动用户-移动服务-上下文3维模型进行降维得到移动用户-移动服务2维模型,最后结合传统2维协同过滤算法进行偏好预测和推荐。仿真数据集和公开数据集实验表明,该算法能够用于移动网络服务环境下的用户偏好预测,并且与传统协同过滤相比具有更高的推荐精确度。     

基于细粒化矩阵的嵌入式知识追踪方法

知识追踪（KT）是根据学习者以前的学习记录预测其未来学习状况的技术。近年来,基于深度学习的知识追踪（DKT）发展迅速,旨在动态追踪学生学习状态进而为学生提供个性化的服务。然而,目前的研究忽略了练习题和知识点之间的复杂联系,为了解决这个问题,提出了一种基于细粒化矩阵的嵌入式知识追踪方法模型（FGM-DKT）。从数据集中提取出了学习实体以及其中隐含关系的嵌入式表达,再将其引入到对学习者的知识建模过程中。在两个真实世界公开数据集上的实验结果表明,提出的FGM-DKT可以更准确地评估学生学习状态和知识点掌握程度,与现有知识追踪模型相比具有更高的预测能力。     

多元协同过滤推荐算法在医科类执业资格考试中的应用

本文提出了一种面向医学类执业资格考试应用的多元协同过滤推荐算法,该算法通过构建学生知识点掌握概率模型,综合考虑执业资格知识点的权重、知识点难易程度、错误率多个因素,将多个因素作为权重因子对基于用户的协同过滤推荐算法进行改进,把多元协同过滤推荐算法应用在执业职格考试系统中。该算法使教师根据分析结果掌握学生普遍存在的课程薄弱知识点,及时调整教学内容和教学方法。使学生根据推荐结果精准把握执业资格考试未掌握的相关知识点,及时调整学习侧重点,增强学习的针对性。实验采用医学高职院校2016-2018三年毕业生医师资格考试医学综合考试为实验数据,最终实验结果验证了多元协同过滤推荐算法可以有效提高学生助理执业医师资格考试成绩。     

融合微聚集隐私保护的协同过滤算法研究

现有的k-匿名隐私保护是一种安全有效的隐私保护算法,针对其对背景知识攻击和同质性攻击防范的不足,提出一种基于敏感属性多样性的微聚集隐私保护的协同过滤算法。算法在满足k-匿名的前提下,融入敏感属性的多样性,在微聚集算法中通过设置同一等价类中敏感属性的差异值,来避免敏感属性值过于接近而造成隐私泄露,从而达到保护隐私数据的目的,同时保证推荐的准确性。实验结果表明,该算法既能保证为用户提供高效的个性化推荐,又能够产生安全的信息表。     

“信号与系统”教学中工程实践能力的培养

"信号与系统"课程的理论概念抽象,课堂上直接理解吸收较困难。教师在授课过程中,适当地引入一些工程实践项目与实验环节可以加强学生对基础理论与工程分析思想的理解与掌握,从而更好地提高学生分析问题和解决问题的能力,以实现学以致用,理论联系实际。     

An affective and Web 3.0-based learning environment for a programming language

We present a Web-based environment for learning Java programming that aims to provide adapted and individualized programming instruction to students by using modern learning technologies as a recommender and mining system, an affect recognizer, a sentiment analyzer, and an authoring tool. All these components interact in real time to provide an educational setting where the student learn to develop Java programs. The recommender system is an E-Learning 3.0 software component that recommends new exercises to a student based on the actions (ratings) of previous learners. The affect recognizer analyze pictures of the student to recognize learning-centered emotions (frustration, boredom, engagement, and excitement) that are used to provide personalized instruction. Sentiment text analysis determines the quality of the programming exercises based on the opinions of the students. The authoring tool is used to create new exercises with no programming work. We conducted two evaluations: one evaluation used the Technology Acceptance Model to assess the impact of our software tool on student behavior. The second evaluation calculated the student’s t-test to assess the learning gain after a student used the tool. The results of the evaluations show the students perceived enjoyment and are willing to use the tool. The study also show that students using the tool have a greater learning gain than those who learn using a traditional method.     

On-line classroom visual tracking and quality evaluation by an advanced feature mining technique

With the rapid development of computer vision and multimedia technology, especially the visual tracking technology and network transmission, teacher-centered education is popular nowadays. The shortcomings of the conventional classroom teaching mode by manually student behavior analysis are gradually becoming less effective. Aiming at the main problems existing in the application of classroom teaching video resources in multimedia teaching, in this paper, we proposed an online classroom visual data tracking system, associated with an advanced tracking quality evaluation method based on data mining. Our proposed framework can offer a scientific basis for improving the quality of online education by discovering students’ learning patterns from their online learning data. The evaluation results can effectively demonstrated that the mining of various learning information of students is useful, and obtained the classification rules that affect the learning effect toward students. These clues can be adopted to uncover the learning effect of students and provide individual guidance for students’ learning behaviors. This work can reveal the pattern online classroom image teaching behavior from the perspective of behavior chain. We also noticed the online classroom visual tracking behavior can be divided into several components: selection, presentation, mapping, analysis and collection, as well as the analysis from students facial expression.     

Learning knowledge representation with meta knowledge distillation for single image super-resolution

Although the deep CNN-based super-resolution methods have achieved outstanding performance, their memory cost and computational complexity severely limit their practical employment. Knowledge distillation (KD), which can efficiently transfer knowledge from a cumbersome network (teacher) to a compact network (student), has demonstrated its advantages in some computer vision applications. The representation of knowledge is vital for knowledge transferring and student learning, which is generally defined in hand-crafted manners or uses the intermediate features directly. In this paper, we propose a model-agnostic meta knowledge distillation method under the teacher–student architecture for the single image super-resolution task. It provides a more flexible and accurate way to help teachers transmit knowledge in accordance with the abilities of students via knowledge representation networks (KRNets) with learnable parameters. Specifically, the texture-aware dynamic kernels are generated from local information to decompose the distillation problem into texture-wise supervision for further promoting the recovery quality of high-frequency details. In addition, the KRNets are optimized in a meta-learning manner to ensure the knowledge transferring and the student learning are beneficial to improving the reconstructed quality of the student. Experiments conducted on various single image super-resolution datasets demonstrate that our proposed method outperforms existing defined knowledge representation-related distillation methods and can help super-resolution algorithms achieve better reconstruction quality without introducing any extra inference complexity.     

Hybrid occupation recommendation for adolescents on interest,profile, and behavior

Young people in high school or college make critical decisions regarding what major to study and which career path to pursue. But, many students enter post-secondary education without a clear idea of their major and future career plans. Discovering students’ suitable occupations as early as possible can help them to choose an appropriate vocational learning direction and to build the skills and the abilities for the prospective occupation. For those reasons, students need an automatic counseling system. In order to do this, recommendation methods were employed; it aims to counsel suitable occupation for students, to discover their occupational interests and to guide them to improve their skills. We implemented a hybrid recommendation system called occupation recommendation (OCCREC) that integrates content-based and collaborative filtering methods. We involved three sets of information including student’s profiles, vocational interests, and their behaviors. The student profile contains two types of data, namely, background and interest/hobby retrieved from Facebook. In the experiment, the students from four countries consisted of Mongolia, Sri Lanka, Taiwan, and Thailand used the OCCREC. And, five occupations were shown to the students by using five similarity measures which are Euclidean, Intersection, Cosine, Jaccard, and Pearson. Finally, OCCREC allows students to rate the results accordingly based on user’s satisfied scores and to share their experiences on Facebook.     

基于模糊认知诊断模型的慕课学习成效评估

慕课以“在线、共享、开放、互动”的独有优势,吸引了众多高校学生及社会兴趣学习者。但其高辍学率和低完成率、缺乏有效的质量评价标准及指标体系、学习效果难以保证等问题急需解决。本文引入心理测量领域的认知诊断模型Fuzzy-CDF,基于MOOC平台湖南大学“信号与系统”课程的教育大数据,提出了知识水平层级的学生学习成效评估方法。通过对章节的诊断性评价来反馈学生知识弱项、推荐资源,保障学习效果与学习的顺利进行；采用过程数据来训练认知诊断模型,完成期中与期末的形成性评价并给出学生最终成绩。实验结果表明：与慕课平台简单的加权形成性评价和期末终结性评价方法相比,本文提出的学习成效评估方法能更好地反映学生的真实水平,具有实际应用参考价值。     

A fuzzy rule-based iterative learning control method with application to hydroforming processes

A pressure tracking control of hydroforming processes which is used for precision forming of sheet metals is considered in this paper. In this process, forming pressure of the process needs to be strictly controlled to ensure the high quality of the forming products. However, satisfactory control performance is often difficult to achieve using conventional control methods due to the complexities and the uncertainties of the process. To overcome the problems, a fuzzy rule-based learning control scheme is proposed. In the proposed scheme, a fuzzy rule base constructed from expert knowledge is adopted to generate an appropriate control input which will provide satisfactory pressure tracking performance of the hydroforming process. A series of experiments was performed to show the effectiveness of the proposed control scheme and to investigate the design parameters of the proposed algorithm. The experimental results show that the proposed fuzzy rule-based learning controller can guarantee good tracking performance and, thus, high quality of the products even when only a vague, imprecise and fragmentary knowledge of the process is available.     

行为导向教学在高职网页设计课程中的应用

针对高职学生的职业能力要求,运用行为导向的教学方法,在网页设计课程的教学中,让学生在行动中学习理论知识,做到理论的适度够用,培养学生的自主学习能力和创新精神。     

Deep knowledge tracking algorithm based on forgetting law

Knowledge tracking(KT) algorithm, which can model the cognitive level of learners, is a fundamental artificial intelligence approach to solve the personalized learning problem in the field of education. The recently presented separated self-attentive neural knowledge tracing(SAINT) algorithm has got a great improvement on predictingthe accuracy of students’ answers in comparison with the present other methods.However there is still potential to enhance its performance for it fails to effectively...     

一种基于社会网络技术的教育协作支持系统设计

为了使用互联网技术更好地促进创新教育,为培养学生创新能力提供有利的环境,设计了一个基于社会网络技术的教育协作支持系统,提出了基于社会网络拓扑结构的相似度概念并以此为基础设计了协作推荐算法。协作推荐算法可以根据学生的兴趣推荐合适的合作伙伴、指导老师、专家、项目或者文档资料。该系统帮助学生更好地应用所学知识解决学习项目,达到创新教育的目的。