Using multimedia for e‐learning

This paper reviews 12 research‐based principles for how to design computer‐based multimedia instructional materials to promote academic learning, starting with the multimedia principle (yielding a median effect size of d = 1.67 based on five experimental comparisons), which holds that people learn better from computer‐based instruction containing words and graphics rather than words alone. Principles aimed at reducing extraneous processing (i.e., cognitive processing that is unrelated to the instructional objective) include coherence (d = 0.70), signalling (d = 0.46), redundancy (d = 0.87), spatial contiguity (d = 0.79) and temporal contiguity (d = 1.30). Principles for managing essential processing (i.e., mentally representing the essential material) include segmenting (d = 0.70), pre‐training (d = 0.46) and modality (d = 0.72). Principles for fostering generative processing (i.e., cognitive processing aimed at making sense of the material) include personalization (d = 0.79), voice (d = 0.74) and embodiment (d = 0.36). Some principles have boundary conditions, such as being stronger for low‐ rather than high‐knowledge learners.     

e‐Learning recommender system for a group of learners based on the unified learner profile approach

In the age of information explosion, e‐learning recommender systems (eL_RSs) have emerged as effective information filtering techniques that attempt to provide the most appropriate learning resources for learners while using e‐learning systems. These learners are differentiated on the basis of their learning styles, goals, knowledge levels and others. Several attempts have been made in the past to design eL_RSs to recommend resources to individuals; however, an investigation of recommendations to a group of learners in e‐learning is still in its infancy. In this paper, we focus on the problem of recommending resources to a group of learners rather than to an individual. The major challenge in group recommendation is how to merge the individual preferences of different learners that form a group and extract a pseudo unified learner profile (ULP) that closely reflects the preferences of all learners. Firstly, we propose a profile merging scheme for the ULP by utilizing learning styles, knowledge levels and ratings of learners in a group. Thereafter, a collaborative approach is proposed based on the ULP for effective group recommendations. Experimental results are presented to demonstrate the effectiveness of the proposed group recommendation strategy for e‐learning.     

基于增量学习的混合推荐算法

推荐系统是自适应信息系统中的个性化服务模块,可以根据目标用户的信息需求提供个性化的信息服务。针对传统协作过滤算法存在的用户兴趣描述粒度过大问题,以及稀疏评分矩阵造成相似度计算不准确的问题,提出了一种基于增量学习的混合推荐算法WHHR,该算法通过Widrow-Hoff增量学习构建基于内容的用户模型,并结合协作过滤推荐机制实现评分预测。实验验证了WHHR算法在收敛速度和推荐准确性方面较类似推荐算法有较大提高。     

基于改进Canopy聚类的协同过滤推荐算法

推荐系统通过建立用户和信息产品之间的二元关系,利用用户行为产生的数据挖掘每个用户感兴趣的对象并进行推荐,基于用户的协同过滤是近年来的主流办法,但存在一定局限性：推荐时需要考虑全部用户,而单个用户往往只与少部分用户类似。为了解决这个问题,提出了基于改进Canopy聚类的协同过滤推荐算法,将用户模型数据密度、距离与用户活跃度结合,计算用户数据权值,对用户模型数据进行聚类。由于结合了Canopy的聚类思想,同一用户可以属于不同的类,符合用户可能对多领域感兴趣的情况。最后对每个Canopy中的用户进行相应的推荐,根据聚类结果与用户评分预测用户可能感兴趣的对象。通过在数据集MovieLens和million songs上与对比算法进行MAE、RMSE、NDGG三个指标的比较,验证了该算法能显著提高推荐系统预测与推荐的准确度。     

LDA-CF: 一种混合协同过滤方法

推荐系统是一种克服信息过载的重要工具,其中最流行的方法是协同过滤。该文提出一种结合潜在因素模型和邻域方法的混合协同过滤方法LDA-CF。我们首先将评分矩阵转换成伪文档集合,使用LDA(Latent Dirichlet Allocation)主题模型发现用户和物品潜在因素向量;然后在低维潜在因素空间计算用户和物品相似度;最后采用邻域方法预测未知评分。在MovieLens 100k数据集上的实验表明: 在评分预测任务中,LDA-CF取得的MAE性能指标优于传统的邻域方法。因此,LDA可以有效地从评分矩阵中发现对计算相似度十分有用的用户和物品低维特征表示,在一定程度上缓解了数据稀疏问题。     

基于双重最相关注意力网络的协同过滤推荐算法

基于项目的协同过滤从用户的历史交互项目中学习用户偏好，根据用户的偏好推荐相似的新项目。现有的协同过滤方法认为用户所交互的一组历史项目对用户的影响是相同的，并且将所有历史交互项目在对目标项目作预测时的贡献看作是相同的，导致这些推荐方法的准确性受限。针对上述问题，提出了一种基于双重最相关注意力网络的协同过滤推荐算法，该算法包含两层注意力网络。首先，使用项目级注意力网络为不同历史项目分配不同的权重来捕获用户历史交互项目中最相关的项目；然后，使用项目交互级注意力网络感知不同历史项目与目标项目之间的交互关联度；最后，通过两层注意力网络的使用来同时捕获用户在历史交互项目上和目标项目上的细粒度偏好，从而更好地进行下一步推荐工作。在MovieLens和Pinterest两个真实数据集上进行实验，实验结果表明，所提算法在推荐命中率上与基准模型基于深度学习的项目协同过滤（DeepICF）算法相比分别提升了2.3个百分点和1.5个百分点，验证了该算法在为用户进行个性化推荐上的有效性。     

基于双重最相关注意力网络的协同过滤推荐算法

基于项目的协同过滤从用户的历史交互项目中学习用户偏好,根据用户的偏好推荐相似的新项目。现有的协同过滤方法认为用户所交互的一组历史项目对用户的影响是相同的,并且将所有历史交互项目在对目标项目作预测时的贡献看作是相同的,导致这些推荐方法的准确性受限。针对上述问题,提出了一种基于双重最相关注意力网络的协同过滤推荐算法,该算法包含两层注意力网络。首先,使用项目级注意力网络为不同历史项目分配不同的权重来捕获用户历史交互项目中最相关的项目;然后,使用项目交互级注意力网络感知不同历史项目与目标项目之间的交互关联度;最后,通过两层注意力网络的使用来同时捕获用户在历史交互项目上和目标项目上的细粒度偏好,从而更好地进行下一步推荐工作。在MovieLens和Pinterest两个真实数据集上进行实验,实验结果表明,所提算法在推荐命中率上与基准模型基于深度学习的项目协同过滤（DeepICF）算法相比分别提升了2.3个百分点和1.5个百分点,验证了该算法在为用户进行个性化推荐上的有效性。     

Conversational Collaborative Recommendation – An Experimental Analysis

Traditionally, collaborative recommender systems have been based on a single-shot model of recommendation where a single set of recommendations is generated based on a user’s (past) stored preferences. However, content-based recommender system research has begun to look towards more conversational models of recommendation, where the user is actively engaged in directing search at recommendation time. Such interactions can range from high-level dialogues with the user, possibly in natural language, to more simple interactions where the user is, for example, asked to indicate a preference for one of k suggested items. Importantly, the feedback attained from these interactions can help to differentiate between the user’s long-term stored preferences, and her current (short-term) requirements, which may be quite different. We argue that such interactions can also be beneficial to collaborative recommendation and provide experimental evidence to support this claim.     

考虑项目属性的协同过滤推荐模型

针对传统的基于用户的协同过滤(UCF)模型在相似性度量过程中没有充分考虑项目属性的问题,提出了两种考虑项目属性的协同过滤推荐模型。模型首先对用户评分相似性进行优化;然后从项目属性的角度统计用户关于不同项目的评价次数,获得优化的基于项目属性的用户相似性;最后通过自适应平衡因子协调处理两方面的相似性结果进行项目预测与推荐。实验结果表明,在不同的数据集中,新提出的模型不仅时间花费较为合理,而且评分预测准确性明显提高,平均提高了5%,从而证明了模型在改进用户相似性度量精度方面的有效性。     

基于项目候选集的协同过滤算法

针对协同推荐技术中的数据稀疏性等问题,提出了一种新颖的协同过滤算法。该算法在用户相似的协同过滤基础之上,引入用户之间的非对称影响度和支持度,提出项目候选集的概念。并且,该算法利用项目信息熵对未评分项目集合进行二次修正,得到候选项目集合,进而提高算法的准确率。基于MovieLens和netflix数据集的实验表明,相比目前多个比较流行的算法,该算法在推荐结果的准确率、召回率和F1值上都具有较大的优势,有效地降低了用户评分稀疏性问题带来的负面影响,显著提高了推荐系统的推荐质量。     

Improving the quality of e‐learning: lessons from the eMM

The quality of e‐learning can be defined in many different ways, reflecting different stakeholders and the complexity of the systems and processes used in higher education. These different conceptions of quality can be mutually contradictory and, while politically significant, may also be beyond the direct control or influence of institutional leaders. The e‐learning maturity model (eMM) provides a framework for e‐learning quality improvement that measures the capability of institutions to sustainably engage in e‐learning and visualizes that capability in a way that assists leaders and managers using that information to undertake systematic and targeted improvements in their institution's e‐learning activities. This paper synthesizes the outcomes of multiple international eMM assessments to extract a set of starting points for institutions intending to improve the quality of e‐learning experienced by their students.     

协同过滤推荐中一种改进的信息核提取方法

针对协同过滤推荐算法中存在的可扩展性问题,在原有基于频率（frequency-based,FB）和排名（rank-based,RB）的信息核提取方法的基础上,提出了改进的提取信息核方法IFB（IFrequency-based）和IRB（IRank-based,IRB）,在寻找最相似邻居环节中提出了一个优化集的概念,在优化集上为每个用户寻找最相似的邻居。从实验结果看出,通过该方法能够得到更加准确的推荐结果,有效降低了绝对平均误差（MAE）,同时具有更高的准确率和召回率,推荐效果更优。     

The role of e‐learning coaches in Australian secondary schools

The purpose of this study was to explore the personal experiences of e‐learning coaches working in secondary schools in order to provide information about the role of an e‐learning coach. Seven secondary schools in the outer‐eastern region of Melbourne, Australia, were involved in a 3‐year e‐learning project. Semi‐structured interviews were conducted by an independent researcher with an e‐learning coach from each of the seven schools involved in the research. A number of themes emerged from the interviews, including role confusion, changes in the role over time, the importance of establishing relationships, barriers to the uptake of information and communication technology (ICT) and ICT usage. To facilitate the integration of ICT with teaching practices in the secondary school setting, we recommend that a full‐time position for an on‐site e‐learning coach receiving the total support of school leaders is needed to provide relevant professional development and ongoing support to classroom teachers.     

融合用户自然最近邻的协同过滤推荐算法

现有的基于近邻的协同过滤推荐方法如基于KNN、基于K-means的协同过滤推荐常用来预测用户评分,但该方法确定邻居个数K非常困难且推荐准确率不高,难以达到理想推荐效果。从选择邻居用户这一角度出发,提出一种融合用户自然最近邻的协同过滤推荐算法（Collaborative Filtering recommendation integrating user-centric Natural Nearest Neighbor,CF3N）,该算法首先自适应地寻找目标用户的自然最近邻居集,再融合目标用户的自然最近邻居集与活动近邻用户集,使用融合后得到的邻居集合预测目标用户评分。实验使用了MovieLens数据集,以RMSE和MAE为评测标准,比较CF3N、CF-KNN与INS-CF算法,结果显示在电影领域该算法的推荐准确率有显著提高。     

Generalization of recommender systems: Collaborative filtering extended to groups of users and restricted to groups of items

In this paper we present a collaborative filtering method which opens up the possibilities of traditional collaborative filtering in two aspects: (1) it enables joint recommendations to groups of users and (2) it enables the recommendations to be restricted to items similar to a set of reference items. By way of example, a group of four friends could request joint recommendations of films similar to “Avatar” or “Titanic”. In the paper, using experiments, we show that the traditional approach of collaborative filtering does not satisfactorily resolve the new possibilities contemplated; we also provide a detailed formulation of the method proposed and an extensive set of experiments and comparative results which show the superiority of designed collaborative filtering compared to traditional collaborative filtering in: (a) number of recommendations obtained, (b) quality of the predictions, (c) quality of the recommendations. The experiments have been carried out on the databases Movielens and Netflix.     

基于单值中智集的协同过滤推荐算法

为缓解推荐系统中用户模糊评价带来的推荐准确性低的问题,构建基于单值中智集的协同过滤推荐模型.首先,构建用户—项目评分矩阵,划分用户评分等级,并将用户评分按照单值中智计算公式转换得到评分对应的真值隶属度、不确定值隶属度、非真值隶属度.然后,引入极端评价计算公式,将其作为极端评分惩罚系数,得到基于单值中智数分数的相似度计算公式.最后,结合杰卡德相关系数得到最终用户相似度.基于单值中智集的协同过滤推荐方法在公开数据集MovieLens上比较验证,实验结果发现融合单值中智集的方法在RMSE、MAE评价指标上均比其他方法有2％～3％的提升,能够有效提高推荐精度,更好地处理模糊问题.     

协同过滤推荐系统综述

随着互联网和信息计算的飞速发展,衍生了海量数据,我们已经进入信息爆炸的时代。网络中各种信息量的指数型增长导致用户想要从大量信息中找到自己需要的信息变得越来越困难,信息过载问题日益突出。推荐系统在缓解信息过载问题中起着非常重要的作用,该方法通过研究用户的兴趣偏好进行个性化计算,由系统发现用户兴趣进而引导用户发现自己的信息需求。目前,推荐系统已经成为产业界和学术界关注、研究的热点问题,应用领域十分广泛。在电子商务、会话推荐、文章推荐、智慧医疗等多个领域都有所应用。传统的推荐算法主要包括基于内容的推荐、协同过滤推荐以及混合推荐。其中,协同过滤推荐是推荐系统中应用最广泛最成功的技术之一。该方法利用用户或物品间的相似度以及历史行为数据对目标用户进行推荐,因此存在用户冷启动和项目冷启动问题。此外,随着信息量的急剧增长,传统协同过滤推荐系统面对数据的快速增长会遇到严重的数据稀疏性问题以及可扩展性问题。为了缓解甚至解决这些问题,推荐系统研究人员进行了大量的工作。近年来,为了提高推荐效果、提升用户满意度,学者们开始关注推荐系统的多样性问题以及可解释性等问题。由于深度学习方法可以通过发现数据中用户和项目之间的非线性关系从而学习一个有效的特征表示,因此越来越受到推荐系统研究人员的关注。目前的工作主要是利用评分数据、社交网络信息以及其他领域信息等辅助信息,结合深度学习、数据挖掘等技术提高推荐效果、提升用户满意度。对此,本文首先对推荐系统以及传统推荐算法进行概述,然后重点介绍协同过滤推荐算法的相关工作。包括协同过滤推荐算法的任务、评价指标、常用数据集以及学者们在解决协同过滤算法存在的问题时所做的工作以及努力。最后提出未来的几个可研究方向。     

Critical issues for e‐learning delivery: what may seem obvious is not always put into practice

Abstract The successful adoption of information and communication technology to enhance learning can be very challenging, requiring a complex blend of technological, pedagogical and organizational components, which may at times require the resolution of contradictory demands and conflicting needs. The research reported in this paper investigated and analysed critical success factors (CSFs) that are required to deliver e‐learning within higher education (HE) courses and programmes. The research design adopted a critical research approach, instantiated by focus group discussions with e‐learning experts drawn from administrative, educational, technology and research domains. The findings revealed that staffing issues, pedagogically sound delivery models and training of both tutors and students cannot be treated as trivial issues and are critical to the success of e‐learning. Furthermore, this research also shows that there is a strong relationship between these factors and inspirational institutional leadership. The findings also suggest that in order to assure the success of e‐learning, this leadership should guarantee the presence of institutional enablers. It is hoped that the CSFs, described and discussed in depth in this paper, will provide a suitable theoretical foundation to underpin the successful delivery of e‐learning within HE.     

协同深度学习推荐算法研究

针对当用户评分较少时,推荐系统由于数据稀疏推荐性能显著降低这一问题,介绍了协同深度学习算法（Collaborative In Deep Learning,CIDL）.本算法首先对大量数据进行深度学习,然后对数据文本进行挖掘提取词汇表,最后对评级（反馈）矩阵进行协同过滤,从而得出对用户的推荐项目.本文使用真实的电影数据进行实验,与另外四种优秀算法进行对比,证明该算法可以真实有效得解决由于数据稀疏使得性能降低的问题,并提高推荐的准确度.     

个性化服务中基于用户聚类的协同过滤推荐

协同过滤技术被成功地应用于个性化推荐系统中，但随着系统规模扩大，它的效能逐渐降低。针对此缺点，使用了基于用户聚类的协同过滤推荐，根据用户评分的相似性对用户聚类，在此基础上搜索目标用户的最近邻居，从而缩小用户的搜索范围。本文还提出将协同过滤推荐分为类内相似系数计算和产生推荐两个阶段，把相似系数的计算放在离线部分，减少在线推荐的计算量，提高实时响应速度。另对聚类算法初始聚类中心的选取也做了改进。