Search Results (30)

In recent years, the increasing adoption of High-Performance Computing (HPC) clusters in scientific research and engineering has exposed challenges such as resource imbalance, node idleness, and overload, which hinder scheduling efficiency. Accurate multidimensional task prediction remains a key bottleneck. To address this, we propose a hybrid prediction model that integrates Informer, Long Short-Term Memory (LSTM), and Graph Neural Networks (GNN), enhanced by a hierarchical attention mechanism combining multi-head self-attention and cross-attention. The model captures both long- and short-term temporal dependencies and deep semantic relationships across features. Built on a multitask learning framework, it predicts task execution time, CPU usage, memory, and storage demands with high accuracy. Experiments show prediction accuracies of 89.9%, 87.9%, 86.3%, and 84.3% on these metrics, surpassing baselines like Transformer-XL. The results demonstrate that our approach effectively models complex HPC workload dynamics, offering robust support for intelligent cluster scheduling and holding strong theoretical and practical significance. Full article

In HPC systems, job scheduling plays a critical role in determining resource allocation and task execution order. With the continuous expansion of computing scale and increasing system complexity, modern HPC scheduling faces two major challenges: a massive decision space consisting of tens of thousands of computing nodes and a huge job queue, as well as complex temporal dependencies between jobs and dynamically changing resource states.Traditional heuristic algorithms and basic reinforcement learning methods often struggle to effectively address these challenges in dynamic HPC environments. This study proposes a novel scheduling framework that combines GTrXL with PPO, achieving significant performance improvements through multiple technical innovations. The framework leverages the sequence modeling capabilities of the Transformer architecture and selectively filters relevant historical scheduling information through a dual-gate mechanism, improving long sequence modeling efficiency compared to standard Transformers. The proposed SECT module further enhances resource awareness through dynamic feature recalibration, achieving improved system utilization compared to similar attention mechanisms. Experimental results on multiple datasets (ANL-Intrepid, Alibaba, SDSC-SP2) demonstrate that the proposed components achieve significant performance improvements over baseline PPO implementations. Comprehensive evaluations on synthetic workloads and real HPC trace data show improvements in resource utilization and waiting time, particularly under high-load conditions, while maintaining good robustness across various cluster configurations. Full article

This study analyses social media networks in tourism using graphs focusing on user influence, hashtag patterns, and engagement. This study aims to reveal the structural function of core users, development of hashtags, and interaction patterns that construct tourism discourses. Using NodeXL 2024 for social network visualization and clustering analysis, this study measures centrality, modularity, and geodesic distances for influential user detection, topical dissemination, and engagement pattern identification. The results uncover bridging nodes between different communities, the proliferation of thematic hashtags related to sustainability and cultural heritage, and the role of emotional and visual storytelling in the use of engagement patterns. The theoretical implications also progress SNA application in tourism studies by illuminating aspects of how online discourses coalesce and the effect of SNA on access. In practical terms, this study indicates that destination marketers must consider leveraging key influencers, using strategic types of hashtags, and by monitoring engagement at key times to maximize effective destination marketing and to enhance crisis communication. These contributions notwithstanding, limitations involve the omission of sentiment analysis and the necessity for longitudinal data. By exploring new emerging platforms like TikTok and Instagram, researchers can begin to understand the more relevant trends of digital engagement. The present research offers a data-driven approach for facilitating the significance of integrating social media strategies with network externalities for tourism operators. Full article

Background and Objectives: Schizophrenia is a significant public health issue, and YouTube has become an increasingly popular source of health information. This study aims to assess the quality and validity of YouTube videos about schizophrenia, focusing on the presence of scientific evidence and the role of healthcare professionals in content quality. Methods: A retrospective, cross-sectional observational study was conducted. One hundred videos in Spanish were selected using NodeXL Pro software, based on specific keywords and hashtags. The videos were categorized by content type and assessed using the DISCERN and Global Quality Scale [GQS] tools to evaluate quality and reliability. Results: Only 39% of the videos referenced scientific articles or technical documents. The videos created by healthcare professionals exhibited a higher quality and reliability. Significant differences were found in the DISCERN and GQS scores between the videos presenting personal opinions and those providing scientific information, favoring the latter. Conclusion: There is a prevalence of misinformation about schizophrenia on YouTube. To enhance the educational value of the platform and reduce misinformation risks, involving healthcare professionals in content creation and implementing control mechanisms is essential. Full article

This study explores the potential of social media in improving education, engagement, and mobilization for climate change initiatives. Using the theoretical framework of resource mobilization and methods such as social network analysis (SNA) and bipartite networks, it examines how effective deployment of resources such as information, social capital, and organizational capabilities can help in the progression of collective movements. Social media platforms, particularly YouTube, significantly influences network structures by facilitating resource mobilization and driving essential engagement. This study extracted data from NodeXL and found that YouTube is an effective medium in disseminating climate change information and delivering educational content to a multilingual audience. Additionally, video affordances such as storytelling, audio–visual effects, and concise narratives enhance viewer interest and engagement, increasing resource mobilization effectiveness. This research offers insights into optimizing social media use for effective resource mobilization and engagement in climate change initiatives. Full article

Social media data are increasingly used to assess public opinion dynamics and develop sustainable regional tourism policies. This study explored the changes in travel patterns and preferences in Gangwon Province before, during, and after the implementation of social distancing measures during the COVID-19 pandemic. Five hundred and twenty-six YouTube videos related to travel in Gangwon Province were collected using NodeXL, and content and statistical analyses were conducted on travel regions, main activities, and viewers’ reactions. The main findings indicated that as the intensity of social distancing measures increased, the activity of YouTube video creators also increased, particularly in the East Coast region, compared with other locations such as mountains, rivers, and traditional markets. Viewer engagement was the highest during the implementation of social distancing, showing a considerable interest in beach travel. These results have significant implications for planning safe travel during crises such as COVID-19 and for local governments to promote a responsible travel environment. Full article

With the rapid advancement of autonomous driving and network technologies, future vehicles will function as network nodes, facilitating information transmission. Concurrently, in-vehicle entertainment systems will undergo substantial enhancements. Beyond traditional broadcasting and video playback, future systems will integrate immersive applications featuring 360-degree views and six degrees of freedom (6DoF) capabilities. As autonomous driving technology matures, vehicle passengers will be able to engage in a broader range of entertainment activities while on the move. However, this evolution in video applications will significantly increase bandwidth demand for vehicular networks, potentially leading to bandwidth shortages in congested traffic areas. This paper presents a method for bandwidth allocation for vehicle video applications within the landscape of vehicle-to-everything (V2X) networks. By utilizing a millimeter-wave (mmWave), terahertz (THz) frequency band, and cell-free (CF) extremely large-scale multiple-input multiple-output (XL-MIMO) wireless communication technologies, we provide vehicle passengers with the necessary bandwidth resources. Additionally, we address bandwidth contention issues in congested road segments by incorporating communication methods tailored to the characteristics of vehicular environments. By classifying users and adjusting according to the unique requirements of various multimedia applications, we ensure that real-time applications receive adequate bandwidth. Simulation experiments validate the proposed method’s effectiveness in managing bandwidth allocation for in-vehicle video applications within V2X networks. It increases the available bandwidth during peak hours by 32%, demonstrating its ability to reduce network congestion and ensure smooth playback of various video application types. Full article

Social media platforms allow people to share information, connect, and build networks at an unprecedented scale with positive and negative consequences. Social network analysis (SNA) applies mathematical network and graph theory to visualise information transfer as relational networks of connected nodes. Measuring node connectivity (centrality) permits the identification of ‘influencers’. SNA has been applied to analyse the spread of misinformation on Twitter (1), but to date, no research has examined nutrition networks. Therefore, this study examined the #Nutrition conversations on Twitter utilising SNA and linguistic analyses. English language tweets including ‘#Nutrition’ on 1–21 March 2023 were collected using the SNA tool, NodeXL Pro (Network Overview for Discovery and Exploration in Excel) (2). SNA is a multistep process that calculates graph metrics and develops a network graph to measure the relationships between users. SNA also identifies semantically related words, hashtags, and word pairs and identifies the sentiment of words used, as measured against the Opinion Lexicon (2). The #Nutrition network included 17,129 vertices (users) with 26,809 unique edges (connections); edges with duplicates were merged. The network density was low, suggesting that most users communicate heavily with a small number of users. The average geodesic distance between any two users was 5.26, revealing a dispersed online discussion. SNA identified the top 10 influencers in this network, measured by high betweenness centrality (23,375,543–5,207,998). Influential users were from a mix of accounts including personal, online blogs, and government organisations. High betweenness centrality identified the users with the greatest influence, acting as bridges between network groups and therefore amplifying #Nutrition messages. Sentiment analysis found the discourse was more positive (0.047, 22,218 words) than negative (0.015, 6795 words). Semantic analysis calculated the total words, 468,191, and identified the most frequently used words in the tweets: #nutrition, #health, food, more, nutrition, health, #diet, #healthylifestlye, #fitness, and #food. Social network analysis shows the discourse on Twitter relating to #Nutrition is dispersed without clear polarising views. Semantic analysis showed that ‘health’ was the main topic discussed in relation to nutrition in this network and was most frequently associated with #Nutrition. The narrative was positively framed, as identified through sentiment analysis. Full article

This study analyzes the quality and reliability of videos related to nutrition and cancer on YouTube. Study Design: An observational, retrospective, cross-sectional, time-limited study analyzing activity on the social network YouTube was proposed. Methods: The information from the videos was extracted through an API search tool, using the NodeXL software. The criteria to select the videos on YouTube were the keywords “real food”, “realfood”, and “cancer” and the hashtags #realfood and #cancer were present, videos in English and videos available on 1 December 2022. Results: The DISCERN value in the total number of videos viewed was 2.25 (±0.88) points, indicating low reliability. The videos uploaded by HRU represented only 20.8%. Videos suggesting that the use of foods defined as “real food” could cure cancer without the intervention of any other treatment accounted for 12.5%. Videos that provided external links to scientific/technical evidence verifying the information represented only 13.89% of the total number of videos. Of these videos, 70% corresponded to HRU. The DISCERN value for videos from HRU users was 3.05 (0.88), a value that reflects a good reliability of videos from these users. Conclusions: This study provides information on the content and quality of the videos that we can find on YouTube. We found videos of non-health users who do not base their content on any scientific evidence, with the danger that this entails for the population, but it also highlights that the videos published by HRU have greater reliability and quality, being better perceived by the population, so it is important to encourage healthcare professionals and health institutions to share verified information on YouTube. Full article

Agriculture has observed significant advancements since smart farming technology has been introduced.The Green Movement played an essential role in the evolution of farming methods. The use of smart farming is accelerating at an unprecedented rate because it benefits both farmers and consumers by enabling more effective crop budgeting. The Smart Agriculture domain uses the Internet of Things, which helps farmers to monitor irrigation management, estimate crop yields, and manage plant diseases. Additionally, farmers can learn about environmental trends and, as a result, which crops to cultivate and how to apply fungicides and insecticides. This research article uses the primary and subsidiary keywords related to smart agriculture to query the Scopus database. The query returned 146 research articles related to the keywords inputted, and an analysis of 146 scientific publications, including journal articles, book chapters, and patents, was conducted. Node XL, Gephi, and VOSviewer are open-source tools for visualizing and exploring bibliometric networks. New facets of the data are revealed, facilitating intuitive exploration. The survey includes a bibliometric analysis as well as a word cloud analysis. This analysis focuses on publication types and publication regions, geographical locations, documents by year, subject area, association, and authorship. The research field of IoT in agricultural plant disease detection articles is found to frequently employ English as the language of publication. Full article

Social media represents a powerful tool for disseminating verified health information on topics such as colostomy, and the roles of healthcare professionals and institutions to ensure the veracity of the information conveyed is increasingly relevant. The main objectives of this study were to analyze the roles of these healthcare professionals and institutions in the conversation about colostomy, without being framed in a specific health communication campaign, and to know the use of reliable information in the conversation. The study was carried out by analyzing Twitter messages containing the hashtag “colostomy” and “Chron” between the 1 January and the 30 April 2022. It was conducted using the NodeXL software, focusing on content analysis of tweets and users’ accounts. The results show that accounts with healthcare activity influence the impressions generated on the network (p = 0.018), finding that nurses are the most active healthcare professionals (22.24%) also having a significant effect on the overall network interactions (p = 0.022). In contrast, we found that institutions do not actively participate on the network. We emphasize the responsibility of institutions for health education and the need for professionals to improve communication skills on social networks, but also the need to improve communication skills on social media to support public health campaigns through these increasingly important channels. Full article

Agenda 2030 and Sustainable Development Goals (SDGs) are critical pieces of climate change communication. #FridaysForFuture (FFF) is one of the movements with the most coverage. This paper analyzes the network structure generated in Twitter by the interactions created by its users about the 23 September 2022 demonstrations, locates the most relevant users in the conversation based on multiple measures of intermediation and centrality of Social Network Analysis (SNA), identifies the most important topics of conversation regarding the #FridaysForFuture movement, and checks if the use of audio-visual content or links associated with the messages have a direct influence on the engagement. The NodeXL pro program was used for data collection and the different structures were represented using the Social Network Analysis method (SNA). Thanks to this methodology, the most relevant centrality measures were calculated: eigenvector centrality, betweenness centrality as relative measures, and the levels of indegree and outdegree as absolute measures. The network generated by the hashtag #FridaysforFuture consisted of a total of 12,136 users, who interacted on a total of 37,007 occasions. The type of action on the Twitter social network was distributed in five categories: 16,420 retweets, 14,866 mentions in retweets, 3151 mentions, 1584 tweets, and 986 replies. It is concluded that the number of communities is large and geographically distributed around the world, and the most successful accounts are so because of their relevance to those communities; the action of bots is tangible and is not demonized by the platform; some users can achieve virality without being influencers; the three languages that stood out are English, French, and German; and climate activism generates more engagement from users than the usual Twitter engagement average. Full article

Social media data reveal patterns of knowledge, attitudes, and behaviours of users on a range of topics. This study analysed 4398 tweets gathered between 17 January 2022 and 3 February 2022 related to ecosystem services, using the keyword and hashtag “ecosystem services”. The Microsoft Excel plugin, NodeXL was used for social and semantic network analysis. The results reveal a loosely dense network in which information is conveyed slowly, with homogeneous, medium-sized subgroups typical of the community cluster structure. Citizens, NGOs, and governmental administrations emerged as the main gatekeepers of information in the network. Various semantic themes emerged such as the protection of natural capital for the sustainable production of ecosystem services; nature-based solutions to protect human structures and wellbeing against natural hazards; socio-ecological systems as the interaction between human beings and the environment; focus on specific services such as the storage of atmospheric CO₂ and the provision of food. In conclusion, the perception of social users of the role of ecosystem services can help policymakers and forest managers to outline and implement efficient forest management strategies and plans. Full article

This paper proposes a methodology for sentiment analysis with emphasis on the emotional aspects of people visiting the Herculaneum Archaeological Park in Italy during the period of the COVID-19 pandemic. The methodology provides a valuable means of continuous feedback on perceived risk of the site. A semantic analysis on Twitter text messages provided input to the risk management team with which they could respond immediately mitigating any apparent risk and reducing the perceived risk. A two-stage approach was adopted to prune a massively large dataset from Twitter. In the first phase, a social network analysis and visualisation tool NodeXL was used to determine the most recurrent words, which was achieved using polarity. This resulted in a suitable subset. In the second phase, the subset was subjected to sentiment and emotion mapping by survey participants. This led to a hybrid approach of using automation for pruning datasets from social media and using a human approach to sentiment and emotion analysis. Whilst suffering from COVID-19, equally, people suffered due to loneliness from isolation dictated by the World Health Organisation. The work revealed that despite such conditions, people’s sentiments demonstrated a positive effect from the online discussions on the Herculaneum site. Full article

Background: Melanoma is the third most common cause of cancer and the deadliest form of skin cancer among 17–39 year-olds in the United States. Melanoma is a critical public health issue with a substantial economic burden. Cases and associated burdens, however, could be prevented with a greater awareness of, and interventions related to, skin cancer and melanoma-related preventive behaviors. In fact, as social media use is close to ubiquitous, it represents a potential communication modality. However, more research is needed to understand the current state of melanoma-related information exchanged between Twitter users. This study aimed to understand the different types of users controlling the melanoma-related information diffusion and conversation themes on Twitter. Methods: Tweets (n = 692) were imported from Twitter between 1 and 31 May 2021 using the Twitter public API; and uploaded to NodeXL to conduct a social network analysis. Results: Health professionals and organizations with medical backgrounds were the main content producers, disseminators, and top influencers. However, information diffusion is slow and uneven among users. Additionally, conversations lacked a focus on preventive behaviors. Conclusion: Twitter is a potential platform for the targeted outreach of individuals in melanoma awareness campaigns. This study provides insights maximizing the effectiveness of Twitter as a communication modality. Our findings can help guide the development of customized content and interventions during melanoma awareness campaigns. Full article