Search

Sina Weibo has evolved into a daily social tool for people, yet effectively leveraging its data for sentiment analysis remains a challenging task due to the presence of information beyond text, such as emojis or images. In this paper, we propose an attention graph convolutional network (AGCN) for fine-grained sentiment classification of Weibo posts. Utilizing an attention network based on cosine similarity, the rich emotional information embedded in emoji features interacts with the textual content, effectively enhancing the capability to represent emotions in the text. Leveraging the characteristics of attention networks to construct a graph structure effectively enables graph convolutional networks to capture higher-order relationships between words in textual features. This approach addresses the challenge of extracting sentiment tendencies from Weibo comments. Experimental results on public data sets demonstrate the effectiveness of AGCNs.

Coxsackievirus B3 (CVB3) is the pathogen causing hand, foot and mouth disease (HFMD), which manifests across a spectrum of clinical severity, ranging from mild to severe. However, traditional CVB3-infected mice studies have predominantly resulted in viral myocarditis and pancreatitis, failing to replicate HFMD symptoms. This has limited our understanding of CVB3 virus-host interaction. Meanwhile, the different clinical manifestations animal models could effectively accelerate the CVB3 novel therapies and vaccines development. Syrian hamsters and rhesus monkeys have been widely used to evaluate many enteroviruses, but CVB3 has not been reported systematically. In this study, we have first tested the susceptibility of Syrian hamsters to CVB3 infection via different routes. Our research discovered that both intraperitoneal injection and nasal drip were effective for CVB3 to infect Syrian hamsters, resulting in characteristic HFMD symptoms, nasopharyngeal colonization, and acute severe pathological injury. Notably, the nasal drip group led to a longer detoxification cycle and more severe pathological damage. In the subsequent study, CVB3 infected rhesus monkeys through nasal drips also presented with HFMD symptoms, detoxification, serum antibody conversion, viral nucleic acids and antigens, and the specific organs pathological damage such as the heart. Surprisingly, no significant differences in myocardial enzyme levels were observed. And the clinical manifestations were akin to those of common, mild infections. In summary, the study established CVB3 severe Syrian hamster and mild rhesus monkey HFMD models, which laid a foundation for understanding the pathogenesis, pre-trial prevention and evaluation and post-exposure intervention.

• A novel hybrid modeling framework for forecasting large number of time series. • Utilizing the meta-ensemble technique to combine the local and global forecasts. • Both point and interval multi-step ahead of forecasting can be generated concurrently. • Achieve superior performance than state-of-art models in load forecasting applications. Many industrial applications concern the forecasting of large quantities of related time series. This paper presents a novel hybrid modeling framework, named as Hybrid Deep Meta-Ensemble Networks (HDME-Nets), which combines local and global forecasts using meta-ensemble technology. The proposed framework can be used to generate multiple steps ahead of point and interval forecasts for a large number of time series concurrently. Our proposed framework is composed of four modules: a set of local forecasters that model each time series individually, a global forecaster that captures cross-sectional patterns with data pooling, a feature learner that extracts features from each time series in a supervised fashion, and a meta-combiner that combines the local and global forecasts according to the extracted features. The local forecasters are fitted firstly, and the other three modules are integrated seamlessly into one neural network, which is then jointly trained with a common objective measured with a custom loss function. Through testing two public available electric utility load datasets, we find that the proposed method can achieve improved forecasting performance compared against some state-of-the-art methods. [ABSTRACT FROM AUTHOR]

The Chinese government has announced a trial programme to provide for private purchase of new battery-operated electric vehicles (EVs) and for plug-in hybrids in five cities. We investigate the potential impact of these subsidies and charging facilities on demand for EVs, using data from a survey of potential car buyers in China. Building on the understanding of factors and incentives that would likely encourage households to adopt EVs would help to improve policy interventions. Data collection is based on experimental design and stated choice methods through an Internet survey. Choice alternatives include a conventional gasoline, a plug-in hybrid and a pure electrical vehicle. In addition to subsidy and charging facilities, we also investigate the impact of common vehicle attributes such as purchase price, cruising range, refuelling time and the socio-demographic effect. Applying a multinomial logit (MNL) model, we find that subsidies would significantly encourage households to choose a plug-in hybrid or an EV. Compared to pure EVs, people in China are more willing to accept plug-in hybrids at present. Compared to the price factors, charging facilities are less of a concern when households consider the adoption of an EV. Willingness-to-pay and market share simulation are computed based on the estimated parameters for further analysis.

Haemophilus influenzaeis a common commensal organism of the human upper respiratory tract and an important cause of human disease. No data on H influenzaecarriage rate has been carried out on the Qinghai-Tibet Plateau of China. This study aims to present the H influenzaecarriage rate and influencing factors of H influenzaein healthy children <15 years of age in Qinghai Province, an area located on the Qinghai-Tibet Plateau in China. Oropharyngeal swabs for the detection of H influenzaeDNA were collected between September and October 2019. Taqman real-time polymerase chain reaction was used to detect the nucleic acids from the oropharyngeal swabs. Self-designed questionnaires were used to investigate the related information among this group of children. A number of 284 children were enrolled in this study. The carriage rate of H influenzaewas 44.7%. The carriage rate in cities was 47.5%, in rural areas was 21.9%, and in pastoral areas was 52.8%. The carriage rate was found to be higher among children of minority ethnic groups than those of Han ethnicity (55.6% vs 38.1%). H influenzaecarriage rate was influenced by tobacco smoke exposure (adjusted odds ratio [aOR] = 2.31, 95% CI [confidence interval]: 1.14–4.70), having siblings <5 years of age (aOR = 2.36, 95% CI: 1.21–4.59), respiratory infections during the last 30 days (aOR = 2.37, 95% CI: 1.11–5.06), and parent/guardian education level (aOR = 0.08, 95% CI: 0.02–0.27). H influenzaewas highly prevalent in healthy children in Qinghai Province, especially among children of minority ethnicities and those living in pastoral areas. Tobacco smoke exposure, having siblings <5 years of age, and respiratory infections during the last 30 days were risk factors for H influenzaecarriage. Parents or guardians having education levels of college or higher was a protective factor for H influenzaecarriage. It is of critical importance that the government take effective measures to reduce the carriage rate and the occurrence of H influenzaerelated diseases in susceptible populations.

The nickel lining on a Raney day: Highly selective catalytic hydrogenation of biphenyl (BP) to cyclohexylbenzene (CHB) was achieved by using skeletal Ni prepared from rapidly quenched Ni–Al alloys ribbons, affording 100 % conversion and 99.4 % selectivity in a one‐pot reaction.

Community detection algorithms play an important role in revealing and analyzing complex network structures. However, when these algorithms are put into practice, privacy concerns often arise, especially when community nodes wish to safeguard their identities from exposure. This interaction between network analysis and privacy issues leads to what we refer to as the "community hiding problem." In the context of concealing specific target communities, popular methods primarily focus on preventing community detection algorithms from accurately identifying the entire original community, resulting in partial concealment. This study formally addresses the challenge of achieving complete concealment of target communities. We introduce key metrics such as escape scores, dispersion scores, and hiding scores to precisely define the problem. Additionally, we design a metric, denoted as M-value, to evaluate the effectiveness of concealing target communities from multiple perspectives. To tackle this challenge, we employ a genetic algorithm that leverages previous knowledge to maximize the concealment of the target community while minimizing changes to link connections. We validate our approach through extensive experiments with various real-world datasets, demonstrating that our algorithm outperforms several state-of-the-art baseline algorithms across multiple metrics. Visual representations of our method confirm its ability to effectively hide target communities while minimally affecting the broader network's community structure, strengthening the inherent effectiveness of our community hiding approach. Furthermore, we assess the adaptability of our modified network with different community detection algorithms, consistently demonstrating effective concealment even when these algorithms are extended. This emphasizes the robustness and generality of our proposed approach in various algorithmic scenarios. [ABSTRACT FROM AUTHOR]