Guang Yu

An analysis of sleep complaints on Sina Weibo

This study explores sleep complaints on Sina Weibo to gain insights into social networking about mental health. A random sample (nź=ź1000) of insomnia-related postings was coded and analyzed to investigate the themes and symptoms expressed in postings. The most common theme mentioned is the disclosure of insomnia. The difficulty with sleep initialization is the most common symptom revealed in postings. Besides, the prevalence of insomnia is higher in certain areas and the people who may suffer from insomnia tend to be active from midnight to noon. Our findings can be used to detect and provide help for those Weibo users who may suffer from insomnia.

Identifying princes of Sleeping Beauty - knowledge mapping in discovering princes

Sleeping Beauty is a typical phenomenon of delayed recognition in scientific discovery. She was unnoticed for a long time, and suddenly found by researchers and cited a lot. This is known as waking-up Sleeping Beauty. The one who arouse the Sleeping Beauty is called prince. In order to discover the waking-up mechanism of Sleeping Beauty, we collect seven classic Sleeping Beautys citing and cited data, and use co-citation cluster analysis method to analyze core citation circle in citation network by CITESPACE. At the same time, we take direct and indirect citations of the Sleeping Beauty into account to identify who was the true prince (he terminated the slumber of the Sleeping Beauty and made a significant contribution to the scientific development). It is a new approach to study the phenomenon of Sleeping Beauty even delayed recognition and identify waking-up mechanism.

Disclosure Pattern of Self-Labeled People Living with HIV/AIDS on Chinese Social Networking Site: An Exploratory Study

HIV/AIDS is an important public health issue in China. The number of people living with HIV/AIDS (PLWHA) has been increasing since the introduction of highly active antiretroviral therapy. PLWHAs life quality is becoming an important issue, with lack of research in China. In this study, a group of PLWHA (n = 663) was identified using HIV/AIDS relevant usernames on a Chinese social networking site (Weibo) to study their daily living situations. We found that more than 99.10% of PLWHA were male, among whom 90.80% turned out to be homosexual. They had significantly more fans and followees, but fewer postings compared to the general population. The mean age of the PLWHA was 28.96 (SD = 5.05) years old, and southwest and northwest China had a high prevalence of HIV/AIDS. In addition, PLWHAs postings were coded and we found that more than half of the postings (n = 769, 51.03%) contained strong emotions. Less than one-fifth of the postings were directly related to HIV/AIDS topics (n = 269, 17.85%), while seeking emotional support, such as looking for stable partnership, was ranked as the first priority of support seeking. In summary, we found that the majority of the self-labeled PLWHA were likely to be men who have sex with men. They used Weibo to share their daily life events and seek emotional support. Implications for promoting HIV/AIDS education and prevention through Chinese social networking sites were also discussed.

A new method for early detection of mass concern about public health issues

During a public health crisis, risk communication professionals and other risk managers need timely and reliable information about the public reaction as soon as possible in order to effectively carry out their responsibilities. Based on a data-set of microblog news posts, this article presents a method that can forecast to what extent news about a public health issue will be disseminated. The study makes advances in rapidly tracking citizens’ reaction towards public health issues by monitoring news media on microblog sites. The findings show that the proposed method can detect early on the intensity of social reaction on a public health issue as well as provide alert signals. The method can also complement existing risk detection systems and help in the design of other powerful risk analytics tools.

Characterizing Depression Issues on Sina Weibo

The prevalence of depression has increased significantly over the past few years both in developed and developing countries. However, many people with symptoms of depression still remain untreated or undiagnosed. Social media may be a tool to help researchers and clinicians to identify and support individuals who experience depression. More than 394,000,000 postings were collected from China’s most popular social media website, Sina Weibo. 1000 randomly selected depression-related postings was coded and analyzed to learn the themes of these postings, and a text classifier was built to identify the postings indicating depression. The identified depressed users were compared with the general population on demographic characteristics, diurnal patterns, and patterns of emoticon usage. We found that disclosure of depression was the most popular theme; depression displayers were more engaged with social media compared to non-depression displayers, the depression postings showed geographical variations, depression displayers tended to be active during periods of leisure and sleep, and depression displayers used negative emoticons more frequently than non-depression displayers. This study offers a broad picture of depression references on China’s social media, which may be cost effectively developed to detect and help individuals who may suffer from depression disorders.

Recognition of college students from Weibo with deep neural networks

Classification of college students is a key to conduct further research on students. In this paper, we collect a set of samples and build deep neural network classifiers to recognize them. We also analyze the experiences and behaviors of the college students on Weibo. Firstly, we manually label 1502 student users and 1498 non-college students. Then, the data about their posts are crawled from Weibo to be transformed into input vectors by feature engineering techniques. Finally, classifiers are built based on two deep learning algorithms, including stacked autoencoders and deep belief network. Experimental results show that deep neural networks performs better than other machine learning algorithms and the classification of the college students can achieve a very high accuracy.

An Analysis of Anxiety-Related Postings on Sina Weibo

This study examines anxiety-related postings on Sina Weibo to gain insight into social networking about mental health. The themes of a random sample of anxiety-related postings (n = 1000) were assessed. The disclosure of anxiety was the most common theme. The prevalence of anxiety was higher in certain areas where the economy is stronger than others, and the people living there suffered from more stress. Users who talked about feeling anxious tended to be more active on social media during leisure hours and less active during work hours. Our findings may be developed to detect and help individuals who may suffer from anxiety disorders at a low cost.

Predicting non-performing loan of business bank by multiple classifier fusion algorithms

Abstract This paper uses multiple classifier fusion methods to predict the non-performing loans (NPL) of business bank. Both macroeconomic and bank-specific variables are collected to form the feature set firstly. Based on selected features, the study applies bagging and AdaBoost algorithms, which are described in this paper as two different method of multiple classifier fusion, to build prediction models. To contrast the classification performances, several basic strong classifiers such as decision tree, k nearest neighbors and support vector machine (SVM) are also adopted. In this experiment, non-performing loans data with 96 features and 10415 instances of a business bank is collected. F-mean and the Area under the ROC Curve (AUC) are considered as metrics of classification performances. The results illustrate that multiple classifier fusion algorithms outperform single basic classifier. Furthermore, the AdaBoost method performs much better than bagging method in processing NPL data of business bank.

Deep learning-based target customer position extraction on social network

In this paper, we extract the target customer attributes and analysis the characteristics of their interests. We classify the accounts into, for example, three data sets, the real estate, healthy parenting and sports. we extract the target customer attributes via deep learning method to study that attributes and build a classification model which is helpful for merchants to find the target customers and make the marketing strategies on social network. We use deep learning method by studying a nonlinear network structure, to achieve complex function approximation and characterization of the input data distribution. We show the strong ability of a few sample concentrated study the data and essential characteristics. The experimental results also show that the DBN outperforms better than the Naive Bayes classifier.