Online Social Capital: Mood, Topical and Psycholinguistic Analysis

Social media provides rich sources of personal information and community interaction which can be linked to aspect of mental health. In this paper we investigate manifest properties of textual messages, including latent topics, psycholinguistic features, and authors’ mood, of a large corpus of blog posts, to analyze the aspect of social capital in social media communities. Using data collected from Live Journal, we find that bloggers with lower social capital have fewer positive moods and more negative moods than those with higher social capital. It is also found that people with low social capital have more randommood swings over time than the people with high social capital. Significant differences are found between low and high social capital groups when characterized by a set of latent topics and psycholinguistic features derived from blogposts, suggesting discriminative features, proved to be useful for classification tasks. Good prediction is achieved when classifying among social capital groups using topic and linguistic features, with linguistic features are found to have greater predictive power than latent topics. The significance of our work lies in the importance of online social capital to potential construction of automatic healthcare monitoring systems. We further establish the link between mood and social capital in online communities, suggesting the foundation of new systems to monitor online mental well-being. Introduction The concept of social capital was introduced in sociology (Bourdieu 1986; Coleman 1988; Putnam 2000). Like other forms of capitals, social capital can be defined by the capacity of facilitating productive activity. In this view, physical capital and human capital can be determined through the value of tools/machines or skills/capabilities respectively (Coleman 1988); meanwhile, social capital is valued by the resources associated with interpersonal relations that facilitate actions (Coleman 1988). For example, (Coleman 1988) noted that a group with trustworthiness, a measure of social capital, has higher chance to accomplish than another without that. Similarly, (Bourdieu 1986) defined social capital as the actual or potential resources a network of relationship possesses. (Putnam 2000) further divided social capital into bonding and bridging based on whether emotional support is provided (bonding), e.g., between family members and close Copyright c © 2013, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. friends (strong ties), or not (bridging), e.g., among acquaintances (weak ties). Generally, social capital includes social inclusion, social participation and social support (VicHealth 2010). Social inclusion implies access for example, older people may have less physical access to social communities. Social participation describes how much a person engages with a community. And, social support implies how much support one can get from the community. The theory of social capital has been spread into several fields, including economics, education and health. For example, according to (Putnam 2001), joining a community might increase one’s social capital and subsequently reduce his chance of fatality in the next year by a half; (Coleman 1988) found that the value of social capital students gained from home and outside is positively correlated with the probability of dropping out of high school. In the age of the Internet, social media genres, such as blogs, social networking sites and forums, have become a popular venue for people to gather, emerging a new form of social capital – digital social capital. This concept has attracted considerable attention, such as investigations into the impact of the new media for the production of social capital (Beaudoin and Tao 2007; Shah, Kwak, and Holbert 2001; Wellman et al. 2001). Insights from such analysis have wide applications, from sociology where the web is viewed as a very large community sensor, to business where e-commerce is now the norm. This study examines a large corpus of blog posts to analyze the effect of online social capital a user potentially possess on mood, topics and psycholinguistic styles conveyed in the messages he creates. Continuing with the core components of social participation and support, we define two categories of LOW and HIGH social capital. We have 12 cohorts to study 2 categories of social capital (LOW and HIGH), and in each, 3 measures of social participation (number of communities joined, number of posts written and number of comments made) and 3 of social support (number of friends, number of comments received and number of followers). Using posts made by 60,000 users in Live Journal over 10 years, we explore across all measures of social participation and social support: (a) distribution of mood and mood swings over LOW and HIGH social capital; (b) distribution of latent topics of interest in LOW and HIGH social capital; and (c) distribution of psycholinguistic features exProceedings of the Seventh International AAAI Conference on Weblogs and Social Media