Community Member Retrieval on Social Media using Textual Information

TL;DR

This paper proposes an unsupervised user embedding method based on re-identification to detect community membership on social media using only text features, especially effective with limited positive examples.

Contribution

It introduces a novel unsupervised proxy task for learning user embeddings that improves community detection accuracy with minimal labeled data.

Findings

Embeddings outperform common unsupervised representations in community detection.

The method is effective across 16 different social media communities.

User re-identification as a proxy task enhances community membership identification.

Abstract

This paper addresses the problem of community membership detection using only text features in a scenario where a small number of positive labeled examples defines the community. The solution introduces an unsupervised proxy task for learning user embeddings: user re-identification. Experiments with 16 different communities show that the resulting embeddings are more effective for community membership identification than common unsupervised representations.