突破智慧教育:基于图学习的课程推荐系统

近年来,随着互联网技术的迅猛发展,以慕课(MOOC)为代表的在线教育平台得到广泛普及.为助力“因材施教”的个性化智慧教育,以推荐算法为代表的人工智能技术受到了学术界与工业界的普遍关注.虽然在电子商务等领域获得了成功应用,但推荐算法与在线教育融合时仍面临严峻挑战:现有算法对隐式交互数据的挖掘不充足,推荐背后的知识指导作用不明显,面向实践的推荐系统软件有缺失.对此,设计了一套面向工业化场景的智慧课程推荐系统:(1)提出基于图卷积神经网络的推荐引擎,将“用户-课程”隐式交互数据建模为异构图;(2)将课程知识信息融入“用户-课程”异构图,深入挖掘了“用户-课程-知识”关联关系;(3)设计了高效的在线推荐系统,实现了“预处理-召回-离线排序-在线推荐-结果融合”的多段流水线原型,不仅能够快速响应课程推荐请求,更能有效缓解推荐算法落地的最大障碍——冷启动问题.最后,基于真实课程学习平台数据集,以对比实验表明了离线推荐引擎相比其他主流推荐算法的先进性,并基于两个典型用例分析验证了在线推荐系统面临工业场景需求的可用性.     

Sharing and cultivating tacit knowledge in an online learning environment

Research on knowledge cultivation often focuses on explicit forms of knowledge. However, knowledge can also take a tacit form—a form that is often difficult or impossible to tease out, even when it is considered critical in an educational context. A review of the literature revealed that few studies have examined tacit knowledge issues in online learning environments. The purpose of this study was to develop a greater understanding of the conditions and processes that help promote the sharing or cultivation of tacit knowledge in a formal online course setting. Using naturalistic inquiry as the methodology of this study, an online graduate business course offered at a private, non-profit United States-based university was purposively selected as the research site. The study found that the online course encouraged processes and created conditions consistent with Nonaka‘s model of knowledge creation and the concept of ba (or shared context)—encouraging students to share, and to construct knowledge through socialization, externalization, combination, and internalization. The results suggest that purposefully developing a ba-like environment may be a useful approach to facilitating online learning, creating a strong potential to support learning processes necessary for students to cultivate tacit knowledge.     

A study of social participation and knowledge sharing in the teachers' online professional community of practice

To facilitate professional development of teachers in the online context, the online community of practice (CoPs) has become an important platform in which individuals with similar interests or common goals get together to share their resources, develop working strategies, solve problems, and improve individual as well as organizational performance. In this study, we have collected self-reported knowledge-sharing behaviors from 321 members of the largest online professional CoP of teachers in Taiwan. The results show that closer connections among online CoP members can lead to greater recognition of and altruism towards others. Moreover, performance expectation and self-efficacy belief play essential roles in knowledge-sharing participation. Thus, the development of social relationships among online teacher members helps them obtain potential resources and reliable support through their social network. Also, teachers' membership in the online professional CoP fosters a prosocial attitude that heightens their willingness to share useful resources and solve other members' problems, both emotionally and instrumentally. Consequently, knowledge-sharing behaviors, in terms of knowledge giving and knowing receiving, are significantly predicted by prosocial commitment and performance expectation respectively. The implications to both research and practice are provided in this paper.     

Understanding Perceived Trust to Reduce Regret

Trust is fundamental for promoting the use of online services, such as e‐commerce or e‐health. Understanding how users perceive trust online is a precondition to create trustworthy marketplaces. In this article, we present a domain‐independent general trust perception model that helps us to understand how users make online trust decisions and how we can help them in making the right decisions, which minimize future regret. We also present the results of a user study describing the weight that different factors in the model (e.g., security, look&feel, and privacy) have on perceived trust. The study identifies the existence of a positive correlation between the user's knowledge and the importance placed on factors such as security and privacy. This indicates that the impact factors as security and privacy have on perceived trust is higher in users with higher knowledge.     

Learners’ navigation behavior identification based on trace analysis

Identifying learners’ behaviors and learning preferences or styles in a Web-based learning environment is crucial for organizing the tracking and specifying how and when assistance is needed. Moreover, it helps online course designers to adapt the learning material in a way that guarantees individualized learning, and helps learners to acquire meta-cognitive knowledge. The goal of this research is to identify learners’ behaviors and learning styles automatically during training sessions, based on trace analysis. In this paper, we focus on the identification of learners’ behaviors through our system: Indicators for the Deduction of Learning Styles. We shall first present our trace analysis approach. Then, we shall propose a ‘navigation type’ indicator to analyze learners’ behaviors and we shall define a method for calculating it. To this end, we shall build a decision tree based on semantic assumptions and tests. To validate our approach, and improve the proposed calculation method, we shall present and discuss the results of two experiments that we conducted.     

An investigation of the validity of course‐based online assessment methods: The role of computer‐related attitudes and assessment mode preferences

The use of e‐assessment methods raises important concerns regarding the reliability and validity of these methods. Potential threats to validity include mode effects and the possible influence of computer‐related attitudes. Although numerous studies have now investigated the validity of online assessments in noncourse‐based contexts, few studies have addressed this issue in an educational context. The present study helps fill this research gap by investigating whether university students' computer‐related attitudes and assessment mode preferences were related to performance on a course‐based online assessment task. Overall, students' attitudes and preferences bore no greater relationships to performance on the online than offline module assessment tasks. This provides support for the validity of course‐based online assessment methods and should help alleviate educators' concerns and encourage more widespread adoption of these methods, helping address the issue of their slow uptake to date. Suggestions for follow‐up studies to corroborate and extend the current findings are offered.     

Toward a Complexity of Online Learning: Learners in Online First-Year Writing

In response to the growing presence of online first-year writing courses, this paper describes a case study of two online first-year writing courses and addresses the questions: What do students in an online first-year writing course perceive as good study habits, and what helps them succeed? Data includes surveys, online discussions, course management statistics, and selected interviews. The study is supported by social cognitive theory described by psychologist Albert Bandura; this methodology allows for examination of internal, external, and behavioral characteristics of participating students. Results of the study indicate that students who rated themselves as making good use of study time also succeeded in the course. Insights from students include information about study activities, management of study time, access to technology, and attitudes about online courses. A surprising result of the study was that students did not consider communication with peers as a productive study activity, despite a deliberate attempt by instructors to build peer interaction into the course. Yet students also reported high levels of engagement and positive attitudes about online learning. The social cognitive lens provides helpful insights about these complex findings by examining the external, internal, and behavioral aspects of online first-year writing students in this study.     

Analyzing collaborative interactions: divergence,shared understanding and construction of knowledge

One of the most important facets of collaborative learning is the interaction between individual and collaborative learning activities – between divergent perspectives and shared knowledge building. Individuals bring divergent ideas into a collaborative environment. While individuals bring their own unique knowledge and perspectives, the second important aspect of collaborative learning is how they move from seemingly divergent perspectives to collaborative knowledge building. This is clearly a social process among group members who could adopt various strategies for resolving differences including asserting dominance, acquiescing, or some form of reciprocal sense making. An important aspect of collaborative learning is the move from assimilation to construction, i.e., creating new understandings based on the discussions that they have had. Documenting this change from divergence to collaborative knowledge building to possible construction is therefore important in understanding the nature the collaborative interactions. In this paper we discuss our analysis of the process of collaborative interactions based on three dimensions – divergence of ideas, collaborative knowledge building and construction. Our aim was to document as well as to understand how collaborative interactions develop over time: whether students raise new issues (ideas) more frequently as they become more familiar with the discussion and discussants, and whether shared knowledge building becomes richer over time, and subsequent evidence that students were able to construct their own understanding based on their interactions with others. Our analyses were conducted in the context of an online graduate course conducted using the learning environment that we designed, CoDE, (Constructivist, Distributed learning Environment). In this paper, we will first describe the design of CoDE. We will then describe a study in which CoDE was used to offer an online graduate course in learning theories. We then discuss our analyses of both individual and collaborative learning as it progressed through the duration of the course.     

Effects of an online problem based learning course on content knowledge acquisition and critical thinking skills

This study investigated how the online problem based learning (PBL) approach employed in an online learning environment influenced undergraduate students’ critical thinking skills (CTS) and content knowledge acquisition. The pretest–posttest control group design was used in the study. The subjects included the students who were enrolled at the Department of Primary School Mathematics Teaching in Anadolu University Education Faculty. Subjects attended to Computer II course in 2008 spring. Experiment group attended the online PBL course whereas the control group attended the online instructor-led course. Each group consisted of 20 students. Data collection tools consisted of a multiple choice content knowledge acquisition scale and the Watson–Glaser critical thinking skills test. The results of two-way mixed design ANOVA indicated that learning in the online PBL group did not have a significant effect on the content knowledge acquisition scores. It was also revealed that learning in the online PBL group had a significant effect on increasing the critical thinking skills.     

On the design of online synchronous assessments in a synchronous cyber classroom

Online learning has been rapidly developing in the last decade. However, there is very little literature available about the actual adoption of online synchronous assessment approaches and any guidelines for effective assessment design and implementation. This paper aims at designing and evaluating the possibility of applying online synchronous assessments in a formal credit course. Relying on the four dimensions of knowledge taxonomy, this paper describes four different online synchronous assessments for conducting a formal mid‐term exam, including synchronous quiz, synchronous practicum, synchronous essay and synchronous oral assessments. The designs were implemented and evaluated in a synchronous cyber classroom for an online credit course. This study identified the challenges and potential issues while conducting the online synchronous assessments, including (1) the extent of monitoring and cheating; (2) the lack of suitable software tools specifically for supporting online synchronous assessments; and (3) the need for a variety of online synchronous assessment methods for different subject matter. This paper concludes with some possible solutions, such as a short time limit for answering questions and a well analysed need of a software tool, to address all the issues.     

Online Crowdsourcing Campaigns: Bottom-Up versus Top-Down Process Model

ABSTRACT

When a crowd’s motivations are not triggered, they may not necessarily commit their best efforts, even if they have the knowledge to answer an open call. Drawing on the incentive theory, we introduce a top-down process model for an online crowdsourcing campaign that addresses the crowd’s motivations. This model is in contrast to the traditional bottom-up process model, where the crowd self-selects an open call based on their knowledge. We adopt a longitudinal case study method and examine two online crowdsourcing campaigns that represent both models. The findings suggest that the campaign that follows the top-down model generated high-quality ideas, while the bottom-up case was considered a failure. We further enrich the top-down model by developing a four-stage guidance model that addresses the crowd’s differing motivations in each stage. This research contributes to the crowdsourcing literature and helps better attract the qualifying crowd, thereby leading to greater campaign success likelihood.     

Automated Integration of External Databases: A Knowledge-Based Approach to Enhancing Rule-Based Expert Systems

Expert system applications in the biomedical domain have long been hampered by the difficulty inherent in maintaining and extending large knowledge bases. We have developed a knowledge-based method for automatically augmenting such knowledge bases. The method consists of automatically integrating data contained in commercially available, external, online databases with data contained in an expert system's knowledge base. We have built a prototype system, named DBX, using this technique to augment an expert system's knowledge base as a decision support aid and as a bibliographic retrieval tool. In this paper, we describe this prototype system in detail, illustrate its use, and discuss the lessons we have learned in its implementation.     

Attitudes, beliefs, and attendance in a hybrid course

Since online learning technology, such as streaming video, was introduced to the college distance-learning environment, there has been a need to study the attitudes of college students toward the use of this modality in hybrid courses. Understanding students’ attitudes toward using online materials and the impact on class attendance is not only essential to ascertain how effectively the online content is delivered, it also helps teaching faculty prepare online courses and education programs. This paper studies the attitudes of college students’ about online materials in a hybrid upper division communications class. Student attitudes towards combining distance-learning techniques with traditional lecture tended to be positive. Students held the belief that hybrid courses have a negative impact on attendance, but they did not self-report an actual impact. Students do believe that they benefit from this technology, but the belief is strongest in those who are most computer/Internet literate. How these attitudes/beliefs impact the student in the context of a rural culture will be discussed.     

Reducing techno-anxiety in high school teachers by improving their ICT problem-solving skills

Teachers need to continuously update their information and communication technologies (ICT) knowledge, but they are usually not trained to deal with the problems arising from their use. In fact, studies in the literature report techno-anxiety (i.e. unpleasant physiological activation and discomfort due to present or future use of ICT) in teachers. Thus, the goal of this action research is to study if teachers’ techno-anxiety can be reduced by increasing their ability to solve technological problems. An inter-subject experiment has been carried out with 46 teachers. High school teachers were chosen because they are digital immigrants, while at the moment of this research their students are digital natives (born around year 2000). Since we could not find any specific training for teachers to increase their resolution skills of technological problems, in order to apply the treatment for our study, we have designed and deployed an online course about ICT problem-solving skills based on the 70/20/10 model for learning and development. Results show the success of the course when it comes to increasing the ICT problem-solving skills and to reducing techno-anxiety.     

A Software Engineering Project Course for Undergraduate Students

Software engineering(SE) courses are confronted with predicaments how to cultivate students' engineering concept and how to abridge the distance between the academy and industry.The "Computing and Software Engineering,III(CSE III)" helps students mimic a real-world software development and broaden realworld SE view.With the spiral model as the main development model,the CSE III course includes not only the SE theory knowledge,but also quick learning,team cooperation,development management and risk management skills which meet the real-world industry requirements.Our contribution is providing an introductory course for multiple dimensions of knowledge,creating a real-world software development circumstance for students under quantitative control and designing a systematic and practical evaluation mechanism which stimulates the interests of students.CSE III has been executed from 2009 and obtains a valuable teaching and learning effects among our teachers,TAs and students;and we add refinements to the course according to feedbacks of participants.This paper will describe the design and practice of the CSE III course.     

A dynamic analysis of the interplay between asynchronous and synchronous communication in online learning: The impact of motivation

With the increased affordances of synchronous communication tools, more opportunities for online learning to resemble face‐to‐face settings have recently become available. However, synchronous communication does not afford as much time for reflection as asynchronous communication. Therefore, a combination of synchronous and asynchronous communication in e‐learning would seem desirable to optimally support learner engagement and the quality of student learning. It is still an open question though, how to best design online learning with a blend of synchronous and asynchronous communication opportunities over time. Few studies have investigated the relationship between learners' actual use of synchronous and asynchronous communication over time. Therefore, this study addressed that relationship in an online course (N = 110), taking into account student motivation, and employing a dynamic inter‐temporal perspective. In line with our assumptions, we found some support for the expected association between autonomous motivation and engagement in asynchronous and synchronous communication, be it restricted primarily to the first course period. Also, positive relations between engagement in synchronous and asynchronous communication were found, with the strongest influence from using asynchronous to synchronous communication. This study adds to the knowledge base needed to develop guidelines on how synchronous communication can be combined with asynchronous learning.     

面向对象知识库的设计

面向对象技术的成熟及其在知识库系统中的应用，赋予了知识库一些新的概念。我们就此进行了较为深入的研究，并对知识表达、推理等作了一些探讨，采用基于面向对象技术的原型－代理机制定义和设计了知识对象，实现了知识共享，并通过对象间的消息传递形成多个推理机场境，从而进行知识的推理。本课题初步实现了一个具有一定推理辅助决策能力的原型知识库系统。     

Exploring and mitigating social loafing in online communities

The motivation to share members’ knowledge is critical to an online community’s survival and success. Previous research has established that knowledge sharing intentions are based on group cohesion. Several studies also suggested that social loafing behavior will seriously corrode group cohesion. Therefore, social loafing is a key obstacle to fostering online community development. Although substantial studies have been performed on the critical factors that affect social loafing in the learning group, those on online communities are still lacking. By integrating two perspectives, social capital and perceived risk, a richer understanding of social loafing behavior can be gained. In the research model, social ties and perceived risk have been driven by anonymity, offline activities, knowledge quality, and media richness. Social ties and perceived risk are hypothesized to affect social loafing in the online community, which, in turn, is hypothesized as negatively affecting group cohesion. Data collected from 323 online users in online communities provide support for the proposed model. The study shows that social loafing is a significant negative predictor of the users’ group cohesion. The study also shows that social ties and perceived risk are important components of social loafing. Anonymity, offline activities, knowledge quality, and media richness all have strong effects on social ties and perceived risk in the online community. Implications for theory and practice and suggestions for future research are discussed.     

Learning and best practices for learning in open-source software communities

This research is about participants who use open-source software (OSS) discussion forums for learning. Learning in online communities of education as well as non-education-related online communities has been studied under the lens of social learning theory and situated learning for a long time. In this research, we draw parallels among these two types of communities and explore what can be learned from open-source software communities about online learning. Thematic network analysis was used to code the qualitative data from the open-ended questions in the survey and the interviews. The results indicate that learning in online open-source software communities encompasses much more than just learning about the software being discussed. 283 Open-source forum participants were surveyed, and 21 were interviewed to develop an understanding of the challenges to learning in these communities as well as to identify the practices that promote learning. Identifying these practices helps to understand online learning and enables the integration of best practices into online education.     

Fake news detection and classification using hybrid BiLSTM and self-attention model

Living in the twenty-first century, from shopping to reading news articles everything has changed, everything has become online. Anyone can access most of everything with a single touch from a cell phone. Internet is the new normal, everyone is very much attached to it. Reading news online is something very common among people of all age groups, thousands of articles are being published on various online media portals online every hour. These articles are not necessarily genuine always, sometimes false information is written knowingly and sometimes knowingly. It is very much needed to keep these articles away from the users. Many kinds of research have been conducted using traditional mathematical models and sequential neural networks to detect this fraud news online. In most of these studies, the news is being analysed in a unidirectional way. Therefore, a need of changing current mechanisms is required to increases the accuracy of false news detection. In this paper, we propose a Bi-LSTM based (Bidirectional long short term memory) deep learning approach by adding self-attention on top of it. This helps in developing a higher clarity, which is the most challenging part of the deep learning paradigm. The classification result demonstrated that the proposed hybrid deep learning model outperforms existing models with an accuracy score of 98.65%.