Few-shot Learning for Multi-modal Social Media Event Filtering

TL;DR

This paper introduces a graph-based few-shot learning approach for filtering social media data related to events, especially protests, using limited labeled samples and a new multi-modal Portuguese protest dataset.

Contribution

It presents a novel few-shot learning pipeline for event filtering on social media and releases the first multi-modal protest dataset in Portuguese for this task.

Findings

Comparable performance with only 60 labeled samples versus 3100 in full data

First multi-modal protest dataset in Portuguese for event filtering

Effective filtering with minimal labeled data

Abstract

Social media has become an important data source for event analysis. When collecting this type of data, most contain no useful information to a target event. Thus, it is essential to filter out those noisy data at the earliest opportunity for a human expert to perform further inspection. Most existing solutions for event filtering rely on fully supervised methods for training. However, in many real-world scenarios, having access to large number of labeled samples is not possible. To deal with a few labeled sample training problem for event filtering, we propose a graph-based few-shot learning pipeline. We also release the Brazilian Protest Dataset to test our method. To the best of our knowledge, this dataset is the first of its kind in event filtering that focuses on protests in multi-modal social media data, with most of the text in Portuguese. Our experimental results show that our proposed pipeline has comparable performance with only a few labeled samples (60) compared with a fully labeled dataset (3100). To facilitate the research community, we make our dataset and code available at https://github.com/jdnascim/7Set-AL.