Learning to Discover Social Circles in Ego Networks

TLDR

Personal social networks are large and cluttered, and existing manual categorization methods on platforms like Google+, Facebook, and Twitter are laborious and require constant updates, leaving no effective way to organize them. The study introduces a new machine learning task to automatically identify users’ social circles. The authors formulate circle detection as node clustering on a user’s ego‑network and train a model that jointly learns circle memberships, circle‑specific profile similarity metrics, and allows overlapping and hierarchical circles. Experiments on Facebook, Google+, and Twitter data with hand‑labeled ground truth demonstrate that the model accurately identifies social circles.

Abstract

Our personal social networks are big and cluttered, and currently there is no good way to organize them. Social networking sites allow users to manually categorize their friends into social circles (e.g. 'circles' on Google+, and 'lists' on Facebook and Twitter), however they are laborious to construct and must be updated whenever a user's network grows. We define a novel machine learning task of identifying users' social circles. We pose the problem as a node clustering problem on a user's ego-network, a network of connections between her friends. We develop a model for detecting circles that combines network structure as well as user profile information. For each circle we learn its members and the circle-specific user profile similarity metric. Modeling node membership to multiple circles allows us to detect overlapping as well as hierarchically nested circles. Experiments show that our model accurately identifies circles on a diverse set of data from Facebook, Google+, and Twitter for all of which we obtain hand-labeled ground-truth.