CHS: Small: Systematic Comparative and Historical Analysis Framework for Social Movements

点击翻译按钮获取中文摘要

This project will develop methods for studying online social movements that are playing increasingly influential roles. Despite its promise, research using social media data has suffered from a number of shortcomings including lack of representativeness, incompleteness, and the difficulty of identifying relevant outcome variables. Such shortcomings are widely acknowledged, yet their solutions have proven elusive. This project makes intellectual contributions to these issues at both the data collection and data labeling steps. this project will improve our understanding of how online movements are formed, how they evolve over time, how the participants coordinate, and how movements change their participants and the broader public. This study will inform community organizers and citizens of strategies that result in desirable outcomes in civic participation, the designers of online platforms that aim to support such civic engagement about which design choices lead to what type of civic engagement, and journalists that aim to provide accurate narratives of online social movements. The project will also have educational impact, as the datasets will be used in data science courses and will be shared with the broader research community. The main goals of this research are to propose methods for optimizing data collection strategies; present methods for optimizing data labeling; analyze the development, coordination, and success for social movements of varying scales under different data collection and labeling schemes; and model determinants of protest participation and movement success using curated data. At the data collection step, the project will contribute: a systematic evaluation of the social media data collection methods used in related work; the formulation of topology-based and content-based data collection strategies; and the development of methods to perform effective iterative data collection given a set of possible strategies, through modeling to maximize the expected reward earned through a sequence of choices made in a system of learning problems. At the data labeling step, the project will contribute the development of cross-domain text classification methods to perform cost-effective data labeling through crowdsourcing platforms, and new insights about the potential for transfer learning across popular social media platforms.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.