Forecasting Ukrainian Refugee Flows With Organic Data Sources

This paper explores the use of organic data to predict forced migration from Ukraine to five neighboring countries. The study combines Twitter conversations with event and fatality data from the Armed Conflict Location and Event Data Project (ACLED) and develops predictive models of forced displacement. The results suggest that Twitter variables were more important predictors in the first phase of the conflict, while event-based predictors were more important in the second phase.

DeMis: Data-efficient Misinformation Detection using Reinforcement Learning

We propose a novel reinforcement learning framework for misinformation detection on Twitter. We release both code, data and pre-trained models.

Language Models

Inferring #MeToo Experience Tweets Using Classic and Neural Models

We propose pre-trained language models for political Twitter data. We evaluate all models and report results. We release both data and pre-trained models.

Knowledge Enhanced Masked Language Model for Stance Detection

We propose a novel language modeling for stance detection. We release both data and pre-trained models.

Stance Detection on Twitter