MDCC: A Multimodal Dynamic Dataset for Donation-based Crowdfunding Campaigns

Xovee Xu,Jiayang Li,Fan Zhou
DOI: https://doi.org/10.1145/3583780.3615124
2023-01-01
Abstract:Crowdfunding platforms have become pivotal financial support avenues for diverse causes, yet the success rates are surprisingly low. Previous research has largely focused on reward-based crowdfunding, leaving donation-based platforms under-studied. In addition, the roles of multimodal data (e.g., textual descriptions and visual photos) and dynamic elements (e.g., sequences of donations, project updates, and comments) in influencing campaign success have been largely overlooked. This paper introduces MDCC, a Multimodal Dynamic dataset for donation-based Crowdfunding Campaigns, collected from 14,961 projects on GoFundMe, incorporates multimodal project information and captures project dynamics, thus providing a comprehensive tool for analyzing donation-based crowdfunding. The dataset is expected to inspire innovative methodologies and facilitate understanding of project success determinants. Our preliminary experiments demonstrate the significance of multimodal and dynamic crowdfunding data on predicting the success of donation-based projects.
What problem does this paper attempt to address?