LFKT:学习与遗忘融合的深度知识追踪模型

知识追踪任务旨在根据学生历史学习行为实时追踪学生知识水平变化,并且预测学生在未来学习表现.在学生学习过程中,学习行为与遗忘行为相互交织,学生的遗忘行为对知识追踪影响很大.为了准确建模知识追踪中学习与遗忘行为,本文提出了一个兼顾学习与遗忘行为的深度知识追踪模型LFKT.LFKT模型综合考虑了四个影响知识遗忘因素,包括学生重复学习知识点的间隔时间、重复学习知识点的次数、顺序学习间隔时间以及学生对于知识点的掌握程度.结合遗忘因素,LFKT采用深度神经网络,利用学生答题结果作为知识追踪过程中知识掌握程度的间接反馈,建模融合学习与遗忘行为的知识追踪模型.通过在真实在线教育数据集上的实验,与当前知识追踪模型相比,LFKT可以更好地追踪学生知识掌握状态,并具有较好的预测性能.     

Enabling people who are blind to experience science inquiry learning through sound‐based mediation

This paper addresses a central need among people who are blind, access to inquiry‐based science learning materials, which are addressed by few other learning environments that use assistive technologies. In this study, we investigated ways in which learning environments based on sound mediation can support science learning by blind people. We used NetLogo, a multi‐agent programmable modeling environment that is widely used for learning about complex systems. In order to provide blind people with access to such models, we used a component that supports sound‐based mediation. The sound‐based mediation provided real‐time information regarding objects' speed, location, and interactions with other objects. We examined blind people's learning about a chemical system of contained gas particles. The study employs a pre‐test intervention–post‐test design. Four adults participated individually in the study. They achieved most referent‐representation connections; their scientific conceptual knowledge became more specific and aligned with scientific knowledge; and their systems reasoning showed greater discrimination and relation between components. Discussion addresses learning with sound‐based mediation in broader terms and suggests further research into the potential of this unique type of low‐cost learning environment to assist blind people in their science learning.     

Association between objective and subjective assessments of environmental ergonomic factors in manufacturing plants

The association between objective and subjective assessments of environmental ergonomic factors including noise, lighting and heat were conducted in a field study in three manufacturing plants. Data were collected from 130 workstations using questionnaire and physical measurements of the noise (noise dosimetry), lighting (task area illuminance) and heat (wet bulb globe temperature – WBGT) levels. The recommended noise, illuminance and WBGT levels were not met in about half of the workstations surveyed, which was in agreement with low satisfaction levels with the environmental factors in the workplace. A considerable effect of the environmental factors was found on perceived workers' job performance, safety and health. The results from contingency coefficient analysis indicated a relatively good agreement between the measured noise, illuminance and WBGT levels and the workers' perception of these factors. The results suggest that quantitative physical measurements should be supplemented by qualitative subjective assessments to provide more specific and additional details about the environmental conditions in each workplace and consequently to improve workers' satisfaction, job performance, safety and health.Relevance to industryThe findings highlight the importance of environmental ergonomics and have implications for improvements in the design of the workplace to enhance workers' satisfaction, job performance, safety and health on areas where the environmental factors are not favourable. A better understanding of the environmental conditions and their effects in each working environment has the potential for a notable impact on productivity and workers' quality of life.     

融合学习者时序行为和认知水平的个性化学习资源推荐算法

个性化服务是构建智慧学习环境的内在要求和建设要点.为学习环境中的主体（学习者）推送个性化学习资源可以提高学习资源的利用概率,解决在线学习容易产生的迷航问题.通过本体知识库的统一性语义建立学习者和学习资源内部结构特征,设计出一个有效计算两者相关性的推荐算法.算法中引入时间衰减函数来描述学习者学习时的时序特征,导入计算学习者认知水平与学习资源难度的匹配算子以体现学习的循序渐进原则.实验结果表明：所构建的时间函数和匹配算子达到了预期目标,更好地提升了所推荐学习资源的质量和适应性,且算法的时间复杂度能满足实时计算要求.     

Harnessing collective knowledge inherent in tag clouds

Tagging systems represent the conceptual knowledge of a community. We experimentally tested whether people harness this collective knowledge when navigating through the Web. As a within‐factor we manipulated people's prior knowledge (no knowledge vs. prior knowledge that was congruent/incongruent to the collective knowledge inherent in the tags). As between‐factor we manipulated whether people had tag clouds available or not. In line with the Information Foraging Theory and with the Co‐Evolution Model of individual learning and collective knowledge building, we found that people's prior knowledge and tag clouds influenced their navigation. Tags which were congruent with people's prior knowledge had a higher probability of being selected. A knowledge test showed that participants adapted their individual conceptual knowledge to the collective knowledge. This incidental learning shows that people harness collective knowledge just by navigation with tag clouds.     

Designing and building real environmental decision support systems

The complexity of environmental problems makes necessary the development and application of new tools capable of processing not only numerical aspects, but also experience from experts and wide public participation, which are all needed in decision-making processes. Environmental decision support systems (EDSSs) are among the most promising approaches to confront this complexity. The fact that different tools (artificial intelligence techniques, statistical/numerical methods, geographical information systems, and environmental ontologies) can be integrated under different architectures confers EDSSs the ability to confront complex problems, and the capability to support learning and decision-making processes. In this paper, we present our experience, obtained over the last 10 years, in designing and building two real EDSSs, one for wastewater plant supervision, and one for the selection of wastewater treatment systems for communities with less than 2000 inhabitants. The flow diagram followed to build the EDSS is presented for each of the systems, together with a discussion of the tasks involved in each step (problem analysis, data collection and knowledge acquisition, model selection, model implementation, and EDSS validation). In addition, the architecture used is presented, showing how the five levels on which it is based (data gathering, diagnosis, decision support, plans, and actions) have been implemented. Finally, we present our opinion on the research issues that need to be addressed in order to improve the ability of EDSSs to cope with complexity in environmental problems (integration of data and knowledge, improvement of knowledge acquisition methods, new protocols to share and reuse knowledge, development of benchmarks, involvement of end-users), thus increasing our understanding of the environment and contributing to the sustainable development of society.     

Effects of interactivity and instructional scaffolding on learning: Self-regulation in online video-based environments

Online learning often requires learners to be self-directed and engaged. The present study examined students' self-regulatory behaviors in online video-based learning environments. Using an experimental design, this study investigated the effects of a newly designed enhanced video learning environment, which was designed to support or scaffold students' self-regulated or self-directed learning on students' learning behaviors and outcomes. In addition, correspondence between students' self-regulation strategies in traditional learning environments and observed self-regulated learning behaviors in the enhanced video environment were examined. A cross-sectional experimental research design with systematic random assignment of participants to either the control condition (common video) or the experimental condition (enhanced video) was utilized. Undergraduate and graduate students participated in the study (N = 80). Study results indicate that the newly designed enhanced video learning environment was a superior instructional tool than the common video learning environment in terms students' learning performance. In addition, there was correspondence between graduate students' self-reported self-regulation and observed self-regulation, with those high on seeking/learning information and managing their environment/behavior more likely to engage more in interactive note-taking.     

Holistic environmental soil-landscape modeling of soil organic carbon

In environmental soil-landscape modeling (ESLM), the selection of predictive variables is commonly contingent on the researchers' domain expertise on soil–environment processes. This variable selection strategy may suffer bias or even fail in regions where the process knowledge is insufficient. To overcome this problem, this study demonstrates a holistic ESLM framework which consists of five components: model conceptualization, data compilation, process identification, parsimonious model calibration, and model validation. Based on the STEP-AWBH conceptual model, a comprehensive pool of 210 potential environmental variables that exhaustively cover pedogenic and environmental factors was constructed. This was followed by strategic variable selection and development of parsimonious prediction models using machine learning techniques. The all-relevant variable selection successfully identified the major and minor factors relevant to the SOC variation, showing that the major factors important for explaining SOC variation in Florida were vegetation and soil water gradient. Topography and climate showed moderate effects on SOC variation. Parsimonious SOC models developed using four minimal-optimal variable selection techniques and simulated annealing yielded optimal predictive performance with minimal model complexity. The holistic ESLM framework not only provides a new view of selecting and utilizing variables for predicting soil properties but can also assist in identifying the underlying processes of soil-environment systems of interest. Due to the flexibility of the framework to incorporate various types of variable selection and modeling techniques, the holistic environmental modeling strategy can be generalized to other environmental modeling domains for both prediction and process identification.     

The study on integrating WebQuest with mobile learning for environmental education

This study is to demonstrate the impact of different teaching strategies on the learning performance of environmental education using quantitative methods. Students learned about resource recycling and classification through an instructional website based on the teaching tool of WebQuest. There were 103 sixth-grade students participating in this study and broken down into three groups: traditional instruction, traditional instruction with WebQuest and WebQuest instruction with outdoors. The major contribution of this study is the introduction of WebQuest into the outdoor instruction. The results of this study show that using WebQuest in outdoor instruction influences students’ learning performance positively. Two other interesting results are: (1) when WebQuest was used in real situations, students could acquire more knowledge and experiences, and (2) in the learning activity of the experiment, the students accomplished different learning tasks and expressed their own opinions and perspectives, which could foster their critical thinking skills. On the other hands, the students in outdoor situation could be positive to participate in learning activity; furthermore, they could ponder the learning contents by observing the real context and then they began to classify/categorize the resources. These findings will contribute to the development of teaching and learning for government, schools and teachers; for instance, teachers act as assistants or tutors and provide students with others public network resources, including PDAs, smartphones, tablet personal computers or the Internet, to improve their learning in outdoor learning environments, such as campus, museums or zoos.     

Teaching training in a mixed-reality integrated learning environment

In this mixed-method study, we examined the design and potential impact of a mixed-reality integrated learning environment (MILE) in providing the simulated and immersive teaching practice for university teaching assistants. A virtual-reality-based learning platform integrating a Kinect-enabled sensorimotor interface was developed and used by twenty three university teaching assistants. Qualitative and quantitative data on the participants' participation behaviors, engagement, and perceptions were collected via video/screen recording, interview, surveys on teaching self-efficacy and sense of presence, and eye tracking. Results indicated that the MILE reinforced sense of presence and supported the performance of an ample range of virtual teaching tasks/actions with avatar-embodied live gesturing. The environmental fidelity in the mixed-reality learning spaces, the design and arrangement of virtual agents and avatars, and the affordance of embodied gesturing and walking are salient MILE design features that affected participants' sense of presence and their virtual teaching performance.     

Students' online interactive patterns in augmented reality-based inquiry activities

Inquiry learning has been developing for years and many countries have incorporated inquiry learning into the scope of K-12 education. Educators have indicated the importance of engaging students in knowledge-sharing activities during the inquiry learning process. In this study, a location-based augmented reality (AR) environment with a five-step guiding mechanism is developed to guide students to share knowledge in inquiry learning activities. To evaluate the effectiveness of the proposed approach in terms of promoting the knowledge sharing behaviors of students, an experiment has been conducted in an elementary school natural science course. The participants were 57 fourth-grade students from an elementary school in Northern Taiwan, divided into an experimental group of 28 students who learned with the AR-based approach and a control group of 29 students who learned with the conventional in-class mobile learning approach. The students' learning behaviors, including their movements in the real-world environment and interactions with peers, were recorded. Accordingly, the learning patterns and interactions of the two groups were analyzed via lag-sequential analysis and quantitative content analysis. It was found that, in comparison with the conventional inquiry-based mobile learning activity, the AR-based inquiry learning activity is able to engage the students in more interactions for knowledge construction. The findings of this study provide guidance for helping teachers develop effective strategies and learning designs for conducting inquiry-based learning activities.     

Workers' education

Health education is one of the most popular weapons against low back pain and other work-related musculoskeletal disorders.

It is used for both primary and secondary prevention to teach healthy life and work habits and to improve body mechanics in order to promote and maintain health. It is assumed that with a few hours of teaching, people adopt new working postures and movements, make ergonomic changes in their environment, and start exercising.

The most common teaching method is lecturing, with visual aids. The practical session is often short.

In evaluations shortly after the sessions, participants have been satisfied, their knowledge has increased, and in some cases the number of injuries or sick leaves has decreased. Follow-ups have shown that the health situation has returned to its former state within two years. In health education, aimed at preventing musculoskeletal injuries, the principle should be ‘learning by doing’. Instead of lecturing, the programmes should have a comprehensive approach, which includes the improvement of the live situation and environment, and involves the entire community.

Three programmes (the prevention of data entry operators' neck and upper limb disorders, early rehabilitation courses for forest workers, and workers guidelines to self-help) are presented and the teaching methods and contents analysed with special interests in the didactic methods and in participation of workers.     

An environmental evaluation learning apprentice system

In engineering problems with a weak domain theory, such as environmental evaluation, machine learning techniques should bring to bear any existing background knowledge so as to guide the knowledge-acquisition process. This paper proposes an Interactive Inductive Learning System (IILS) which can use background knowledge provided by experts to avoid incorrect or incomplete induced heuristics. Through the specification of guidance relations, the expert can force the rule-induction system to focus on a subset of relevant attributes and training instances. A guidance relation consists of a set of attribute constraints, where each constraint instructs IILS on the role of an attribute for the induction of a concept. IILS has been linked to the database of an Environmental Evaluation Support System (EESS) for the induction of impact estimation and comparison heuristics. The induced heuristics are generalizations of past assessments stored in the database and permit the prediction of impact levels for cases not previously encountered by EESS. The integration of IILS transforms EESS into an environmental impact assessment learning apprentice system capable of acquiring and improving evaluation heuristics through its normal use.     

Mobile Robot Learning by Self-Observation

The aim of the project described in this paper was to investigate robot learning at a most fundamental level. The project focused on the transition between organisms with innate behaviors and organisms that have the most rudimentary capability of learning through their personal interaction with their environment. It was assumed that the innate behaviors gave basic survival competence but no learning ability. By observing the interaction between their innate behaviors and the organism's environment it was reasoned that the organism should be able to learn how to modify its actions in a way that improves its performance. If a learning system is given more information than it requires then, when it is successful, it is difficult to say which pieces of information contribute to the success. For this reason the information available to the learning system was kept to an absolute minimum. In order to provide a practical test of the learning scheme developed in this project, the robot environment EDEN was constructed. Within EDEN a robot's actions influence its internal energy reserves. The environment incorporates sources of energy, and it also involves situations that use additional energy or reduce energy consumption. A successful learning scheme was developed purely based on the recorded history of the robot's interactions with its environment and the knowledge that the robot's innate behavior was reactive. This learning scheme allowed the robot to improve its energy management by exhibiting classical conditioning and a restricted form of operant conditioning.     

A language learning support system using course-centered ontology and its evaluation

This paper presents a course-centered ontology for assisting learning support systems to embody the relations among knowledge points and also among the learning materials for those knowledge points. An “individual-class-individual” ontology design (first an individual-class design, then an innovative design about relations among bottom individuals), was applied to the construction of a course-centered ontology for an existing Japanese grammar course. Furthermore, a customizable language learning support system was built to manipulate the course-centered ontology to provide an interface for the learning objects arrangement which displays the visual representation of knowledge points and their relations. The intention underlying the development of the system is to encourage instructors to orient their teaching materials to specific knowledge points and even directly to relations between knowledge points. With these orientations, the learning support system is able to provide an environment in which learners can readily distinguish between related knowledge points. Finally, based on the result of a preliminary evaluation, a study to explore the impact of learning styles and learning habits on learning performance was conducted to further evaluate our ontology-based learning support system. The results of the study suggest three main points: (a) the experimental students who learned with our system achieved significantly better learning achievement than those who just did self-study with textbooks after studying the same target contents for 60 mins; (b) the learning achievement of experimental group was not related to either their learning style in Sequential/Global dimension or their habit of “learning from comparison”; (c) in terms of the learning perception of experimental group, compared to “Sequential learners”, most “Global learners” had a stronger feeling that the comparison function is useful in improving their learning performance, and the learners who don't habitually “learning from comparison” were more likely to suffer from lack of the attention and feel more pressure than those who do habitually “learning from comparison”.     

Perceptions of built environment and health outcomes for older Chinese in Beijing: A big data approach with street view images and deep learning technique

Built environment attributes have been demonstrated to be associated with various health outcomes. However, most empirical studies have typically focused on objective built environmental measures. Still, perceptions of the built environment also play an important role in health and may complement studies with objective measures. Some built environment attributes, such as liveliness or beauty, are difficult to measure objectively. Traditional methods to assess perceptions of the built environment, such as questionnaires and focus groups, are time-consuming and prone to recall bias. The recent development in machine deep learning techniques and big data of street view images, makes it possible to assess perceptions of the built environment with street view images for a large-scale study area. By using online free Tencent Street View (TSV) images, this study assessed six perceptual attributes of the built environment: wealth, safety, liveliness, depression, bore and beauty. These attributes were associated with both the physical and the mental health outcomes of 1231 older adults in 48 neighborhoods in the Haidian District, Beijing, China. Results show that perceived safety was significantly associated with both the physical and mental health outcomes. Perceived depression and beauty were significant related to older adults' mental health, while perceived wealth, bore and liveliness were significantly related to their physical health. The findings carry important policy implications and hence contribute to the development of healthy cities. It is urgent to improve residents' positive perceptions and decrease their negative perceptions of the built environment, especially in neighborhoods that are highly populated by older adults.     

Hybrid autonomous control for multi mobile robots

Reinforcement learning can be an adaptive and flexible control method for autonomous system. It does not need a priori knowledge; behaviors to accomplish given tasks are obtained automatically by repeating trial and error. However, with increasing complexity of the system, the learning costs are increased exponentially. Thus, application to complex systems, like a many redundant d.o.f. robot and multi-agent system, is very difficult. In the previous works in this field, applications were restricted to simple robots and small multi-agent systems, and because of restricted functions of the simple systems that have less redundancy, effectiveness of reinforcement learning is restricted. In our previous works, we had taken these problems into consideration and had proposed new reinforcement learning algorithm, 'Q-learning with dynamic structuring of exploration space based on GA (QDSEGA)'. Effectiveness of QDSEGA for redundant robots has been demonstrated using a 12-legged robot and a 50-link manipulator. However, previous works on QDSEGA were restricted to redundant robots and it was impossible to apply it to multi mobile robots. In this paper, we extend our previous work on QDSEGA by combining a rule-based distributed control and propose a hybrid autonomous control method for multi mobile robots. To demonstrate the effectiveness of the proposed method, simulations of a transportation task by 10 mobile robots are carried out. As a result, effective behaviors have been obtained.     

Do emotions matter in technology training? Exploring their effects on individual perceptions and willingness to learn

Due to the growing importance of enterprise resource planning (ERP) systems in organizations today, many universities and corporate training programs now incorporate such systems into their curricula using novel tools, such as ERP simulation games. To improve the understanding of individuals' learning patterns, we extend prior cognition based models by incorporating emotions. Moreover, we highlight the role of perceived radicalness, which mediates the relationship between emotions and individuals' willingness to learn. The proposed research model draws from the appraisal tendency framework and includes four distinct classes of emotions: challenge, achievement, loss, and deterrence. The model was tested via a lab experiment in which participants (N = 145) played an ERP simulation game. The hypothesized links were tested using structural equation modeling. Anger and excitement were negatively associated with perceived radicalness. Anxiety was positively associated with perceived radicalness and the positive effect of happiness on perceived radicalness was insignificant. Finally, perceived radicalness was positively related to individuals' willingness to learn. Results indicate that perceived radicalness is an important mechanism via which classes of emotions impact students' ERP learning behaviors. In addition, negative emotions have greater effects on radicalness perceptions compared to positive emotions.     

Online learning of contexts for detecting suspicious behaviors in surveillance videos

One of the main incentives for implementing video-based surveillance systems is the urban security. During the last years, several approaches for automatic detection of suspicious events have been proposed. Those methods usually require a training stage before starting their operation. This means that previous to run time a representative dataset of interest events, that may occur in the future, must be available. Nevertheless, most real surveillance systems lack of that information, so many of those proposals results impractical.In this paper, a context online learning scheme for detecting suspicious behaviors on surveillance videos is proposed. Contextual information, which is inferred from videos of people in a scenario, allows detecting suspicious behaviors before an eventual criminal's final attack occur. The main attribute of the proposed approach is the capacity to start up its operation with a reduced training dataset. By an incremental learning process, which uses new data obtained during the online operation, the proposed scheme improves the performance over time achieving a better adaptation to conditions of each scenario.The proposed scheme was validated on two datasets. The first of them includes threats against a parked truck and its driver. The second testing dataset is composed of night assault scenes recorded in an urban environment. The experimental results demonstrate that the proposed method is able to learn incrementally from a reduced initial dataset, achieving a performance similar to batch-type systems trained with all data simultaneously and outperforming five state-of-the-art algorithms over violence detection.     

Development of an environmental virtual field laboratory

Laboratory exercises, field observations and field trips are a fundamental part of many earth science and environmental science courses. Field observations and field trips can be constrained because of distance, time, expense, scale, safety, or complexity of real-world environments. Our objectives were to develop an environmental virtual field laboratory to study environmental properties and processes that stimulate the higher-order cognitive skills of students. We considered the following criteria for our virtual field laboratory: (i) global access, i.e., web-based implementation; (ii) simulation of a variety of learning mechanisms; (iii) interactivity to engage students; (iv) compartmentalization and hierarchical organizational structure; (v) abstraction of 2D and 3D geographic objects (e.g. soils, terrain) and dynamic ecosystem processes (e.g. water flow) using geostatistics and scientific visualization techniques. Cognitive science was considered during the design of our computer-aided instructional tools to enhance the effectiveness for learning. Our virtual field laboratory mimicked the students' learning processes that operate during real field trips and/or field observations; and provided students with a simulation environment to study environmental processes in space and time that cannot be provided on a real field trip. We implemented the following learning mechanisms: (i) exploration-based learning; (ii) analogy-based learning; (iii) science inquiry learning; (iv) abstraction-based learning. To engage students in our environmental virtual field laboratory, we implemented multiple interactivity functions including the exploration of 3D models and adaptive selective simulations. We used Virtual Reality Modeling Language, Java, Java Script, and External Authoring Interface to develop the environmental virtual field laboratory for a 42-ha flatwood site in Florida for which extensive datasets existed. Our digital learning environment offers potential to enhance existing on-campus courses and/or distance education courses.