Description

We are living in the era of social networks, where people throughout the world are connected and organized by multiple social networks. The views revealed by different social networks may vary according to the different services they offer. They are complimentary to each other and comprehensively characterize a specific user from different perspectives. As compared to the scare knowledge conveyed by a single source, appropriate aggregation of multiple social networks offers us a better opportunity for deep user understanding.



Currently, we focus on research on multiple social network learning regarding user profiling. In particular, we studied an application scenario: volunteerism tendency prediction. In modern society, volunteers are extremely crucial to NPOs to sustain their continuing operations. The discovery of users' volunteerism tendency can significantly facilitate the recruitment of volunteers for NPOs, which can save considerable time and money to find the potential volunteers. In this work, we explored three popular social networks: Twitter, Facebook and LinkedIn, as they are representative of a public, private, and professional social network, respectively.



Here we introduce a multi-social network dataset, including:

  • 2,499 users, with 1,079,672 tweets and 1,257,755 followings' profiles. A typical user Twitter profile looks like this.
  • 1,271 users, with 193,757 timelines. A typical user Facebook profile looks like this.
  • 2,499 users, A typical user LinkedIn profile looks like this.



Three types of features extracted from these data, including

  • Demographic Characteristics
  • Linguistic Features
  • LIWC features
  • User topics
  • Contextual topics
  • Behavior-based Features
  • Posting behavior patterns
  • Egocentric network patterns



Features

Twitter Features

Facebook Features

LinkedIn Features



Ground Truth

Ground Truth