Search results

With the continuous growth of online education, the quality issue of online educational videos has become increasingly prominent, causing students in online learning to face the dilemma of knowledge confusion. The existing mechanisms for controlling the quality of online educational videos suffer from subjectivity and low timeliness. Monitoring the quality of online educational videos involves analyzing metadata features and log data, which is an important aspect. With the development of artificial intelligence technology, deep learning techniques with strong predictive capabilities can provide new methods for predicting the quality of online educational videos, effectively overcoming the shortcomings of existing methods. The purpose of this study is to find a deep neural network that can model the dynamic and static features of the video itself, as well as the relationships between videos, to achieve dynamic monitoring of the quality of online educational videos.

The quality of a video cannot be directly measured. According to previous research, the authors use engagement to represent the level of video quality. Engagement is the normalized participation time, which represents the degree to which learners tend to participate in the video. Based on existing public data sets, this study designs an online educational video engagement prediction model based on dynamic graph neural networks (DGNNs). The model is trained based on the video’s static features and dynamic features generated after its release by constructing dynamic graph data. The model includes a spatiotemporal feature extraction layer composed of DGNNs, which can effectively extract the time and space features contained in the video's dynamic graph data. The trained model is used to predict the engagement level of learners with the video on day T after its release, thereby achieving dynamic monitoring of video quality.

Models with spatiotemporal feature extraction layers consisting of four types of DGNNs can accurately predict the engagement level of online educational videos. Of these, the model using the temporal graph convolutional neural network has the smallest prediction error. In dynamic graph construction, using cosine similarity and Euclidean distance functions with reasonable threshold settings can construct a structurally appropriate dynamic graph. In the training of this model, the amount of historical time series data used will affect the model’s predictive performance. The more historical time series data used, the smaller the prediction error of the trained model.

A limitation of this study is that not all video data in the data set was used to construct the dynamic graph due to memory constraints. In addition, the DGNNs used in the spatiotemporal feature extraction layer are relatively conventional.

In this study, the authors propose an online educational video engagement prediction model based on DGNNs, which can achieve the dynamic monitoring of video quality. The model can be applied as part of a video quality monitoring mechanism for various online educational resource platforms.

This study aims to explore the underlying mechanisms that support the resilience of the Chinese hospitality industry during and after the COVID-19 epidemic.

Content analysis was applied to 133 manually collected text articles about COVID-19 responses and strategies.

A two-step learning model (emergency reaction, precautions and prevention stages) was identified in the study. In the emergency reaction step, the primary strategies were related to customers, employees, suppliers and facility/food. In the precautions and prevention step, the strategies were related to customers, employees, suppliers and society/public relations. Multiple stakeholders are discussed in the two circles over a continual process in the learning, reacting and adapting stages.

A gap in the literature is filled by this study, providing a learning model and synthesizing various strategies applied in the hotel sector for multiple stakeholders.

This paper takes the seven overcapacity industries such as the textile industry, electricity and heat, steel, coal, automobile manufacturing, nonferrous metals and petrochemical industry as research objects and proposes a TOPSIS grey relational projection group decision method with mixed multiattributes, which is used for the ranking of the seven industries with overcapacity and provided relevant departments with a basis for decision-making.

First, an evaluation index system from four aspects is established. Secondly, the attributes of linguistic information are converted into two-dimensional interval numbers and triangular fuzzy numbers, and an evaluation matrix is constructed and normalized. This paper uses the AHP method to determine the subjective weights and uses the coefficient of variation method to determine the objective weights. Moreover, this paper sets up the optimization model with the largest comprehensive evaluation value to determine the combined weights. Finally, the TOPSIS grey relational projection method is proposed to calculate the closeness of grey relational projections and to rank them.

This paper analyzes the problem of overcapacity in seven industries with the TOPSIS grey relational projection method. The results show that the four industries of automobile manufacturing, textile, coal and petrochemical are all in serious overcapacity levels, while the three industries of steel, nonferrous metals and electric power are relatively in weak overcapacity level in the three years of 2016–2018. TOPSIS grey relational projection method ranks the overcapacity degree of the seven major overcapacity industries, making the relative overcapacity degree of each industry more clear and providing a reference for the government to formulate targeted policies and measures for each industry.

By using TOPSIS grey relational projection method to evaluate the overcapacity of the seven major overcapacity industries, on the one hand, it makes the relative overcapacity degree of each industry more clear, on the other hand, it can provides the basis for the government and decision-making departments. This helps them promote better the healthy and orderly economic development of the seven major industries and avoid resource waste caused by overcapacity.

This article solves the single evaluation method caused by the limited indicators in the past, combines TOPSIS and the grey relational projection method and applies it to the overcapacity evaluation of the industry, not only applies it to the evaluation of overcapacity for the first time but also involves novel problems and methods, which expands the scope of application of the model.

Walking-aid exoskeletons can assist and protect effectively the group with lower limb muscle strength decline, workers, first responders and military personnel. However, there is almost no united control strategy that can effectively assist daily walking. This paper aims to propose a hybrid oscillators’ (HOs) model to adapt to irregular gait (IG) patterns (frequent alternation between walking and standing or rapid changing of walking speed, etc.) and generate compliant and no-delay assistive torque.

The proposed algorithm, HOs, combines adaptive oscillators (AOs) with phase oscillator through switching assistive mode depending on whether or not the AOs' predicting error of hip joint degree is exceeded our expectation. HOs can compensate for delay by predicting gait phase when in AOs mode. Several treadmill and free walking experiments are designed to test the adaptability and effectiveness of HOs model under IG.

The experimental results show that the assistive strategy based on the HOs is effective under IG patterns, and delay is compensated totally under quasiperiodic gait conditions where a smoother human–robot interaction (HRI) force and the reduction of HRI force peak are observed. Delay compensation is found very effective at improving the performance of the assistive exoskeleton.

A novel algorithm is proposed to improve the adaptability of a walking assist hip exoskeleton in daily walking as well as generate compliant, no-delay assistive torque when converging.

Food festivals are prevalent for those passionate about food experience globally. More importantly, feedback from food reviewers on mass media platforms has been becoming a critical factor in facilitating the decision-making process of tourists in particular cities. Moreover, stimulating local tourism activities, thanks to food festivals, prove advantageous to the well-being of local habitants. The purpose of this paper is to provide readers with a general overview of food festival research trends in tourist cities, as tourism has the potential to contribute to targets in Goals 8, 12 and 14 on sustainable consumption and production and the sustainable use of resources, respectively, (UNWTO: World Tourism Organization).

This study searched and filtered documents from the Scopus and Web of Science databases, as well as used bibliometric analysis and other mathematical and statistical methods, to better understand the food festival research context between 1970 and 2021. The carriers with mathematical and statistical methods. VOSviewer algorithm was used to identify critical input for visualizing bibliometric networks and to create a framework for this academic food festival research.

The findings are primarily related to pre and post-COVID-19 research on food festivals worldwide. Furthermore, using an inductive approach, this paper reveals the impact of food festivals in cities and tourist behaviors. According to the findings, the food festival research trends are about “food festivals,” “slow food festivals” and “local food festivals.” Factor analysis is one of the most common analyses in this type of research. Other studies could use the findings and limitations to select appropriate themes and analysis approaches for their research topics.

Research data sets are mainly from articles that may not account for all actual trends during this pandemic.

This review expects to provide insights into food festivals and help future researchers to recognize several research gaps such as the lack of research on food festival manufacturers and producers or the consistency in visitors' aspect research of quality service, visitors' loyal intentions, satisfaction and culinary experience. The tourism industry can find research trends of food festivals and issues following COVID-19 to find their management styles to fit the context of the post-COVID-19 pandemic, facilitating organizing a safe and effective food festival.

This study aims to empirically examine the relationships among perceived environmental uncertainty (EV), the level of knowledge distance (KD) and the impact of value network on firm performance.

The quantitative analysis is based on data from 243 Chinese companies with engineering, procurement and construction (EPC) business in the context of the COVID-19 pandemic.

The two dimensions of value network [network centrality (NC) and network openness (NO)] have a different impact on firm performance [financial performance (FP) and market performance (MP)]. NC has a positive impact on FP, but not on MP. NO has a positive effect on MP, but not on FP. A reduced KD mediates the relationship between value network and firm performance. Moreover, it fully mediates the relationship between NC and MP, NO and FP. Finally, during the COVID-19 pandemic, only EV has a moderating effect on KD and MP.

This study is limited in terms of data set because it relies on a limited amount of cross-sectional data from one specific country. Therefore, researchers are encouraged to test the proposed propositions further.

The present findings suggest that EPC professionals should pay more attention to the EV, which may be impacted by policy, technology and the economy. This research has actionable implications for the reform of EPC in the construction industry, and practical recommendations for EPC firms to improve their corporate performance.

The results measure the complementary effects of both dimensions of value network (NC and NO) on two distinct aspects of firm performance (MP and FP) and assess the moderating effect of EV and KD in the context of the COVID-19 pandemics.

Flexible eddy current array (FECA) sensor is flexible and light in weight, which has broad application prospects in structural health monitoring. But, the sensor’s sensing channel number is more, increasing the added mass of sensor networks. This paper aims to reduce the sensing channel number by changing the sensing coil layout.

In this paper, FECA sensors with series sensing coil (SSC) layout and interactive sensing coil (ISC) layout are proposed, which reduce the number of sensor’s channels by half. Then, the variation of the output signal of the sensor when the crack expands along both sides of the hole is analyzed by simulation model. Finally, the fatigue crack monitoring experiment is carried out.

For the SSC layout, the simulation results show that the amplitude of each SSC group of the sensor increases when the crack propagates to the left or right. For the ISC layout, when the crack propagates on the right side of bolt hole, the induced voltage of each ISC group decreases. When the crack propagates on the left side of bolt hole, the induced voltage of each ISC group increases. The experiment results are consistent with simulation results, which verifies the correctness of simulation model. Compared with SSC layout, the ISC layout can judge the crack propagation direction. And the crack monitoring accuracy is 1 mm.

The research results provide a certain reference for reducing the number of sensor’s sensing channels. Results of the simulation and experiment show that the ISC layout can judge the crack propagation direction, and the crack monitoring accuracy is 1 mm.

This paper aims to study the influence of different reinforcement methods on crack monitoring characteristics of eddy current array sensors, and the sensors with two different reinforcement methods, SUS304 reinforcement and permalloy reinforcement, are proposed.

First, the finite element model of the sensor is established to analyze the influence of the reinforcement plate’s electromagnetic parameters on the crack identification sensitivity. Then, the crack monitoring accuracy test of sensors with two reinforcement methods is carried out. Finally, the fatigue crack monitoring experiments with bolt tightening torques of 45 and 63 N · m are carried out, respectively.

In this study, it is found that the crack identification sensitivity of the sensor can be improved by increasing the relative permeability of the reinforcement plate. The crack monitoring accuracy of the sensors with two different reinforcement methods is about 1 mm. And the crack identification sensitivity of the sensor reinforced by permalloy reinforcement plate is significantly higher than that of the sensor reinforced by SUS304 reinforcement plate.

The sensor reinforced by reinforcement plate can work normally under the squeezing action of the bolt, and the crack monitoring sensitivity of the sensor can be significantly improved by using the reinforcement plate with high relative permeability.

Excavating valuable outlier information of gray privacy products, the purpose of this study takes the online reviews of women’s underwear as an example, explores the outlier characteristics of online commentary data, and analyzes the online consumer behavior of consumers’ gray privacy products.

This research adopts the social network analysis method to analyze online reviews. Based on the online reviews collected from women’s underwear flagship store Victoria’s Secret at Tmall, this study performs word segmentation and word frequency analysis. Using the fuzzy query method, the research builds the corresponding co-word matrix and conducts co-occurrence analysis to summarize the factors affecting consumers’ purchase behavior of female underwear.

Establishing a formal framework of gray privacy products, this paper confirms the commonalities among consumers with respect to their perceptions of gray privacy products, shows that consumers have high privacy concerns about the disclosure or secondary use of personal private information when shopping gray privacy products, and demonstrates the big difference between online reviews of gray privacy products and their consumer descriptions.

The research lays a solid foundation for future research in gray privacy products. The factors identified in this study provide a practical reference for the continuous improvement of gray privacy products and services.

Lenders in online peer-to-peer (P2P) lending platforms are always non-experts and face severe information asymmetry. This paper aims to achieve the goals of gaining high returns with risk limitations or lowering risks with expected returns for P2P lenders.

This paper used data from a leading online P2P lending platform in America. First, the authors constructed a logistic regression-based credit scoring model and a linear regression-based profit scoring model to predict the default probabilities and profitability of loans. Second, based on the predictions of loan risk and loan return, the authors constructed linear programming model to form the optimal loan portfolio for lenders.

The research results show that compared to a logistic regression-based credit scoring method, the proposed new framework could make more returns for lenders with risks unchanged. Furthermore, compared to a linear regression-based profit scoring method, the proposed new framework could lower risks for lenders without lowering returns. In addition, comparisons with advanced machine learning techniques further validate its superiority.

Unlike previous studies that focus solely on predicting the default probability or profitability of loans, this study considers loan allocation in online P2P lending as an optimization research problem using a new framework based upon modern portfolio theory (MPT). This study may contribute theoretically to the extension of MPT in the specific context of online P2P lending and benefit lenders and platforms to develop more efficient investment tools.