My research sits at the crossroads of computer science and sociology, more popularly known as computational social science. Blending ideas from both these fields, I uncover insights about social life online via large datasets. I have examined credibility in social media, conspiratorial thinking among Twitter anti-vaxxers, strategies for obtaining high-quality crowd annotations, language in crowd-funding campaigns, and gossip in organizational emails.
A Parsimonious Language Model of Social Media Credibility Across Disparate Events
Tanushree Mitra|CSCW 2017|Paper
We present a parsimonious model that maps language cues to perceived levels of credibility. Our results show that certain linguistic categories and their associated phrases are strong predictors surrounding disparate social media events. For example, hedge words and positive emotion words are associated with lower credibility.
Understanding Anti-Vaccination Attitudes in Social Media
Tanushree Mitra|ICWSM 2016|Paper
By using four years of longitudinal data capturing vaccination discussions on Twitter, we find that those with long-term anti-vaccination attitudes manifest conspiratorial thinking and mistrust in government. New adoptees are predisposed to form anti-vaccination attitudes via similar government distrust.
Recovery Amid Pro-Anorexia: Analysis of Recovery in Social Media
By developing a statistical framework using survival analysis, we find that recovery on Tumblr is protracted. Only half of the population shows likelihood of recovery after four years, and a vast minority is not estimated to recover even at the end of six years.
CREDBANK: A Large-scale Social Media Corpus With Associated Credibility Annotations
Tanushree Mitra |ICWSM 2015| Paper| Data
PRESS: New Scientist| The Independent
We present CREDBANK, the first systematically studied large-scale credibility corpus of social media events and their in-situ judgments of accuracy. CREDBANK is based on the real-time tracking of billions of streaming tweets over a period of three months, computational summarizations of those tweets, and intelligently routing tweet streams to human annotators.
Comparing Person-and Process-centric Strategies for Obtaining Quality Data on Amazon Mechanical Turk
Tanushree Mitra|CHI 2015| Paper
Best Paper Honorable Mention
The emergence of crowd-sourced micro labor markets like Amazon Mechanical Turk (AMT) is attractive for researchers who wish to acquire large-scale independent human judgments. Yet acquiring high quality judgments using an online workforce is a challenge. We design and conduct a large, controlled experiment to measure the efficacy of selected strategies for obtaining high quality data annotations from non-experts. Our results point to the advantages of person-oriented strategies over process-oriented strategies.
Modeling Factuality Judgments in Social Media Text
|ACL 2014| Paper
As events unfold, journalists and political commentators use quotes — often indirect — to convey potentially uncertain information and claims from their sources and informants. By obtaining annotations of perceived certainty of quoted statements in Twitter and comparing the ability of linguistic and extra-linguistic features to predict readers’ assessment of the certainty, we find that readers are influenced by linguistic framing devices and do not consider other factors, e.g. sources, journalist.
The Language that Gets People to Give: Phrases that Predict Success on Kickstarter
Tanushree Mitra |CSCW 2014| Paper| Data| Slides
PRESS: Mashable| Washington Post| Forbes| Fast Company| New Scientist
We explore the factors which lead to funding on Kickstarter. Applying natural language methods and statistical analysis techniques to a corpus of crowdfunded projects, we find that the language used in the project has surprising predictive power–accounting for 58.56% of the variance around successful funding. A closer look at the phrases shows they exhibit general persuasion principles.
Analyzing Gossip in Workplace Email
Tanushree Mitra |ACM Newsletter Winter 2013|Paper
Adopting the Enron email dataset and natural language techniques, we find that workplace gossip is common at all levels of the organizational hierarchy, with people most likely to gossip with their peers
Have You Heard?: How Gossip Flows Through Workplace Email
Tanushree Mitra |ICWSM 2012|Paper| Poster
PRESS: Science Daily| Huffington Post| Yahoo News| Business News Daily
Gossip is fundamental to social life. Here, we present the first large-scale study of gossip in CMC, looking at email where someone is mentioned in the message body but not included on the recipient list. We find that gossip emails are often more negative and people have a greater likelihood of sending gossip messages to smaller audiences.
Cost, Precision, and Task Structure in Aggression-Based Arbitration for Minimalist Robot Cooperation
Tanushree Mitra |SAB 2012|Paper
- CS 4472: Design of Online Communities
- CS 8803: Social Computing
- CS4464/6465: Computational Journalism
Training under Tech to Teaching Program
- CETL 8713: Fundamentals of Teaching and Learning
- CETL 8717: Course Design for Higher Education
- Senior Program Committee: ICWSM 2016
- Program Committee: ICWSM 2014-2015, WWW 2015
- Reviewer: CHI 2014-2017, CSCW 2013-2017, WSDM 2014, WWW 2016
- Student Volunteer: ACL 2014, CSCW 2015