Search results

A community demonstrates the unique qualities and relationships between its members that distinguish it from other communities within a network. Network analysis relies heavily on community detection. Despite the traditional spectral clustering and statistical inference methods, deep learning techniques for community detection have grown in popularity due to their ease of processing high-dimensional network data. Graph convolutional neural networks (GCNNs) have received much attention recently and have developed into a potential and ubiquitous method for directly detecting communities on graphs. Inspired by the promising results of graph convolutional networks (GCNs) in analyzing graph structure data, a novel community graph convolutional network (CommunityGCN) as a semi-supervised node classification model has been proposed and compared with recent baseline methods graph attention network (GAT), GCN-based technique for unsupervised community detection and Markov random fields combined with graph convolutional network (MRFasGCN).

This work presents the method for identifying communities that combines the notion of node classification via message passing with the architecture of a semi-supervised graph neural network. Six benchmark datasets, namely, Cora, CiteSeer, ACM, Karate, IMDB and Facebook, have been used in the experimentation.

In the first set of experiments, the scaled normalized average matrix of all neighbor's features including the node itself was obtained, followed by obtaining the weighted average matrix of low-dimensional nodes. In the second set of experiments, the average weighted matrix was forwarded to the GCN with two layers and the activation function for predicting the node class was applied. The results demonstrate that node classification with GCN can improve the performance of identifying communities on graph datasets.

The experiment reveals that the CommunityGCN approach has given better results with accuracy, normalized mutual information, F1 and modularity scores of 91.26, 79.9, 92.58 and 70.5 per cent, respectively, for detecting communities in the graph network, which is much greater than the range of 55.7–87.07 per cent reported in previous literature. Thus, it has been concluded that the GCN with node classification models has improved the accuracy.

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.

Session-based recommendation aims to predict the user's next preference based on the user's recent activities. Although most existing studies consider the global characteristics of items, they only learn the global characteristics of items based on a single connection relationship, which cannot fully capture the complex transformation relationship between items. We believe that multiple relationships between items in learning sessions can improve the performance of session recommendation tasks and the scalability of recommendation models. At the same time, high-quality global features of the item help to explore the potential common preferences of users.

This work proposes a session-based recommendation method with a multi-relation global context–enhanced network to capture this global transition relationship. Specifically, we construct a multi-relation global item graph based on a group of sessions, use a graded attention mechanism to learn different types of connection relations independently and obtain the global feature of the item according to the multi-relation weight.

We did related experiments on three benchmark datasets. The experimental results show that our proposed model is superior to the existing state-of-the-art methods, which verifies the effectiveness of our model.

First, we construct a multi-relation global item graph to learn the complex transition relations of the global context of the item and effectively mine the potential association of items between different sessions. Second, our model effectively improves the scalability of the model by obtaining high-quality item global features and enables some previously unconsidered items to make it onto the candidate list.

Purchase conversion prediction aims to improve user experience and convert visitors into real buyers to drive sales of firms; however, the total conversion rate is low, especially for e-retailers. To date, little is known about how e-retailers can scientifically detect users' intents within a purchase conversion funnel during their ongoing sessions and strategically optimize real-time marketing tactics corresponding to dynamic intent states. This study mainly aims to detect a real-time state of the conversion funnel based on graph theory, which refers to a five-class classification problem in the overt real-time choice decisions (RTCDs)—click, tag-to-wishlist, add-to-cart, remove-from-cart and purchase—during an ongoing session.

The authors propose a novel graph-theoretic framework to detect different states of the conversion funnel by identifying a user's unobserved mindset revealed from their navigation process graph, namely clickstream graph. First, the raw clickstream data are identified into individual sessions based on a 30-min time-out heuristic approach. Then, the authors convert each session into a sequence of temporal item-level clickstream graphs and conduct a temporal graph feature engineering according to the basic, single-, dyadic- and triadic-node and global characteristics. Furthermore, the synthetic minority oversampling technique is adopted to address with the problem of classifying imbalanced data. Finally, the authors train and test the proposed approach with several popular artificial intelligence algorithms.

The graph-theoretic approach validates that users' latent intent states within the conversion funnel can be interpreted as time-varying natures of their online graph footprints. In particular, the experimental results indicate that the graph-theoretic feature-oriented models achieve a substantial improvement of over 27% in line with the macro-average and micro-average area under the precision-recall curve, as compared to the conventional ones. In addition, the top five informative graph features for RTCDs are found to be Transitivity, Edge, Node, Degree and Reciprocity. In view of interpretability, the basic, single-, dyadic- and triadic-node and global characteristics of clickstream graphs have their specific advantages.

The findings suggest that the temporal graph-theoretic approach can form an efficient and powerful AI-based real-time intent detecting decision-support system. Different levels of graph features have their specific interpretability on RTCDs from the perspectives of consumer behavior and psychology, which provides a theoretical basis for the design of computer information systems and the optimization of the ongoing session intervention or recommendation in e-commerce.

To the best of the authors' knowledge, this is the first study to apply clickstream graphs and real-time decision choices in conversion prediction and detection. Most studies have only meditated on a binary classification problem, while this study applies a graph-theoretic approach in a five-class classification problem. In addition, this study constructs temporal item-level graphs to represent the original structure of clickstream session data based on graph theory. The time-varying characteristics of the proposed approach enhance the performance of purchase conversion detection during an ongoing session.

In this chapter we develop a new approach, based on the identification of strategy classes, to study how firms face multiple demands. The procedure that we propose (called Relational Class Analysis) stems from an analysis of the similarity of associative patterns across multiple observable outcomes, which reflect the underlying set of choices firms make to similarly address demands. Empirically, the study of 18 financial and extra-financial performance outcomes for 3,655 firms shows the existence of three main strategic classes. Drawing on our analysis, we redefine strategy as the set of committed decisions undertaken to resolve trade-offs between multiple concurrent objectives and discuss the implications of our approach for eight core questions for strategy and organizational theory.

This paper aims to clarify the resistance degree of group road logistics to flood disaster resilience. The paper measures the resilience of group road logistics by establishing network structure model. The purpose of this study is to improve the resilience of road log.

This paper adopts Delphi method to collect data, interviews mainly flood management experts and supply chain risk management experts, and then analyzes the data through the network structure model combined with interpretative structure model (ISM) and analytical network process (ANP).

The results show that flood frequency and drainage systems are the main factors affecting the resilience of road transport logistics in urban areas. These research results provide useful guidance for the effective planning and design of urban road construction and infrastructure.

However, the main factors affecting the resilience of road transport logistics are likely to change with the development of factors such as climate, economy and environment. Therefore, in future work, the authors' research will focus on the further application of this evaluation method.

The results show that the impact of flooding on the four dimensions of road logistics resilience varies. This shows that in deciding what intervention measures are to be taken to improve the resilience of the road network to flooding, various measures need to be considered.

This paper provides a more scientific analysis of the risk management ability of the road network in the face of floods. In addition, it also provides a useful reference for urban road planners.

This paper addresses a clear need to study how to build models to improve the resilience of road logistics in flood risk.

This paper aims to investigate whether the establishment of commodity futures can effectively hedge systemic risk in the spot network, given the context of financialization in the commodity futures market.

Utilizing industry association data from the Chinese commodity market, the authors identify systemically important commodities based on their importance in the production process using multiple graph analysis methods. Then the authors analyze the effect of listing futures on the systemic risk in the spot market with the staggered difference-in-differences (DID) method.

The findings suggest that futures contracts help reduce systemic risks in the underlying spot network. Systemic risk for a commodity will decrease by approximately 5.7% with the introduction of each corresponding futures contract, since the hedging function of futures reduces the timing behavior of firms in the spot market. Establishing futures contracts for upstream commodities lowers systemic risks for downstream commodities. Energy commodities, such as crude oil and coal, have higher systemic importance, with the energy sector dominating systemic importance, while some chemical commodities also have considerable systemic importance. Meanwhile, the shortest transmission path for risk propagation is composed of the energy industry, chemical industry, agriculture/metal industry and final products.

The paper provides the following policy insights: (1) The role of futures contracts is still positive, and future contracts should be established upstream and at more systemically important nodes in the spot production chain. (2) More attention should be paid to the chemical industry chain, as some chemical commodities are systemically important but do not have corresponding futures contracts. (3) The risk source of the commodity spot market network is the energy industry, and therefore, energy-related commodities should continue to be closely monitored.

The purpose of this paper is to investigate and quantify the impact of product complexity, including architectural complexity, on operator learning, productivity and quality performance in both assembly and disassembly operations. This topic has not been extensively investigated in previous research.

An extensive experimental campaign involving 84 operators was conducted to repeatedly assemble and disassemble six different products of varying complexity to construct productivity and quality learning curves. Data from the experiment were analysed using statistical methods.

The human learning factor of productivity increases superlinearly with the increasing architectural complexity of products, i.e. from centralised to distributed architectures, both in assembly and disassembly, regardless of the level of overall product complexity. On the other hand, the human learning factor of quality performance decreases superlinearly as the architectural complexity of products increases. The intrinsic characteristics of product architecture are the reasons for this difference in learning factor.

The results of the study suggest that considering product complexity, particularly architectural complexity, in the design and planning of manufacturing processes can optimise operator learning, productivity and quality performance, and inform decisions about improving manufacturing operations.

While previous research has focussed on the effects of complexity on process time and defect generation, this study is amongst the first to investigate and quantify the effects of product complexity, including architectural complexity, on operator learning using an extensive experimental campaign.

The purpose of this research is to identify quick, effective and affordable architectural design solutions to improve the health of patients, visitors and staff, in an underfunded community healthcare center (CHC) in Baltimore.

Both qualitative (individual and Focus group interviews) and quantitative methods (space syntax analysis and questionnaire survey) are used to assess the healing environment. The questionnaires cover interior and exterior environment related questions.

The questionnaire identifies three issues related to wayfinding, interior design, and access to natural light. The findings demonstrate that the waiting area, exam room and laboratory need to be upgraded. Incorporation of some other interior components including, wall art, plants, music, signage are also suggested by the respondents.

Very few studies in the USA have been done to understand the effects of architectural design to create a healing environment on the community scale.

Considering the frequency of extreme events, enhancing the global financial system's stability has become crucial. This study aims to investigate the contagion effects of extreme risk events in the international commodity market on China's financial industry. It highlights the significance of comprehending the origins, severity and potential impacts of extreme risks within China's financial market.

This study uses the tail-event driven network risk (TENET) model to construct a tail risk spillover network between China's financial market and the international commodity market. Combining with the characteristics of the network, this study employs an autoregressive distributed lag (ARDL) model to examine the factors influencing systemic risks in China's financial market and to explore the early identification of indicators for systemic risks in China's financial market.

The research reveals a strong tail risk contagion effect between China's financial market and the international commodity market, with a more pronounced impact from the latter to the former. Industrial raw materials, food, metals, oils, livestock and textiles notably influence China's currency market. The systemic risk in China's financial market is driven by systemic risks in the international commodity market and network centrality and can be accurately predicted with the ARDL-error correction model (ECM) model. Based on these, Chinese regulatory authorities can establish a monitoring and early warning mechanism to promptly identify contagion signs, issue timely warnings and adjust regulatory measures.

This study provides new insights into predicting systemic risk in China's financial market by revealing the tail risk spillover network structure between China's financial and international commodity markets.