Weakly Supervised Cross-platform Teenager Detection with Adversarial BERT

TL;DR

This paper introduces a novel adversarial BERT-based framework for cross-platform teenager detection that effectively transfers knowledge from labeled source data to unlabeled target platforms, improving detection accuracy.

Contribution

The paper presents a new weakly supervised cross-platform teenager detection method using adversarial BERT, capable of operating without labeled data on the target platform.

Findings

Significant improvement over baseline models in cross-platform detection accuracy

Effective knowledge transfer from source to target social media platforms

Framework works with limited labeled data from source and none from target

Abstract

Teenager detection is an important case of the age detection task in social media, which aims to detect teenage users to protect them from negative influences. The teenager detection task suffers from the scarcity of labelled data, which exacerbates the ability to perform well across social media platforms. To further research in teenager detection in settings where no labelled data is available for a platform, we propose a novel cross-platform framework based on Adversarial BERT. Our framework can operate with a limited amount of labelled instances from the source platform and with no labelled data from the target platform, transferring knowledge from the source to the target social media. We experiment on four publicly available datasets, obtaining results demonstrating that our framework can significantly improve over competitive baseline models on the cross-platform teenager detection task.