Bin Zhu

Social media is a major platform for opinion sharing. To better understand and exploit opinions on social media, we aim to classify users with opposite opinions on a topic for decision support. Rather than mining text content, we introduce a link-based classification model named Global Consistency Maximization (GCM) that partitions a social network into two classes of users with opposite opinions. Experiments on a Twitter dataset show that: (1) our global approach achieves higher accuracy than two baseline approaches; and (2) link-based classifiers are more robust to small training samples if selected properly.

Senior citizens could greatly be benefited from the social support received from a community (Choi et al. 2014; Goswami et al. 2010). Social support denotes to the interaction/communication with others, verbal or nonverbal, reducing the uncertainty or enhancing the self-perception of in control of one’s own life (Albrecht and Adelman 1987). All participants of online communities are motivated by their desire of seeking social support. And such support occurs when community members form relational links among them and have interactions that intend to help (Heaney and Israel 2002). A network member can receive/send different types of social supports from/to others. Informational support transmits information and provides guidance related to the task/question a community member has (Krause 1986); emotional support expresses understanding, encouragement, empathy affection, affirming, validation, sympathy, caring and concern (House 1981; Wang et al. 2014); companionship or network support gives the recipient a sense of belonging (Keating 2013; Wang et al. 2014); and appraisal support enhances the self-evaluation of the recipient (House 1981). Studies have shown that people are usually motivated by their desire of seeking one or more types of social supports to participate in an online community (Goswami et al. 2010; Kanayama 2003; Pfeil 2007; Pfeil and Zaphiris 2009; Wright 2000; Xie 2008). And such social support can only be acquired during the interaction with others. For senior citizens, even though they can be greatly benefited from the social support received through participation, the obstacles they need to overcome in order to feel engaged could be larger than that of younger people (Charness and Boot 2009; Lee et al. 2011), especially when they come to the community for the first time. They could be easily overwhelmed by the content that has been generated by other existing members, finding it difficult to identify an appropriate member to initiate a meaningful interaction. It therefore is critical for an online community system to help senior participants identify other existing members who are more likely to supply the type of support they are seeking. While many previous studies have uncovered the variety factors, contextual (Pfeil and Zaphiris 2009; Wang et al. 2015; Xie 2008) or individual (Wang et al. 2014, 2015, 2012; Wright 1999), that impact the degree to which a senior citizen receives social support needed from an online community, it remains unclear what the characteristics of existing community members who are more likely to provide a new comer the kind of support, informational, emotional, companionship, or appraisal are. And the answer to this question may have significant academic and practical implications. This study thus proposes to fulfil the gap by utilizing data collected from a senior community website to investigate the links between the characteristics of existing senior members and the amount and the type of support they provided to new comers.