Abstract: Left-leaning social thoughts are not a unitary and coherent theoretical system, and leftists can be divided into divergent groups. Based on inductive qualitative observations, this article proposes a theoretical typology of two dimensions of theoretical resources and position orientations to describe left-wing social thoughts communicated in online space. Empirically, we used a mixed approach, an integration of case observations and big-data analyses of Weibo tweets, to investigate three types of left-leaning social thoughts. The identified left-leaning social thoughts include state-centered leftism, populist leftism, and liberal leftism, which are consistent with the proposed theoretical typology. State-centered leftism features strong support of the state and the current regime and a negative attitude toward the West, populist leftism is characterized by unequivocal affirmation of the revolutionary legacy and support for disadvantaged grassroots, and liberal leftism harbors a grassroots position and a decided affirmation of individual rights. In addition, we used supervised machine learning and social network analysis techniques to identify online communities that harbor the afore-mentioned left-leaning social thoughts and analyzed the interaction patterns within and across communities as well as the evolutions of community structures. We found that during the study period of 2012–2014, the liberal leftists gradually declined and the corresponding communities dissolved; the interactions between populist leftists and state-centered leftists intensified, and the ideational cleavage between these two camps increased the online confrontations. This article demonstrates that the mixed method approach of integrating traditional methods with big-data analytics has enormous potential in the sub-discipline of digital sociology.

Key words: Internet, social thoughts, left wing, right wing, populism, big data, network analysis