This repository contains the implementation and dataset for our work on "Profiling Cryptocurrency Influencers with Few-shot Learning," which is part of the CLEF 2023 Conference and Labs of the Evaluation Forum workshop.
The task is to predict the level of influence that cryptocurrency influencers have on Twitter using a novel few-shot learning approach.
Our team ranked 14th out of 27 teams, successfully outperforming all benchmark models presented at the workshop.
The paper detailing our methodology, experiments, and results can be accessed here at CEUR WS: "UZHatPAN-2023:ProfilingCryptocurrency InfluencersusingEnsembleofLanguageModels".
The trained model that we have used for this project is available for download through this Google Drive link: Few-shot Learning Model.