comp.social papers

A Parsimonious Language Model of Social Media Credibility Across Multiple Events

Tanushree Mitra, Graham Wright & Eric Gilbert | CSCW 2017

We present a parsimonious model that maps language cues to perceived levels of credibility. For example, hedge words and positive emotion words are associated with lower credibility.

What (or Who) Is Public? Privacy Settings and Social Media Content Sharing

Casey Fiesler et. al | CSCW 2017

Our results from survey and qualitative analysis show the user awareness of privacy setting options on Facebook. We also analyzed what (or who) are more likely to be posted publicly.

Popup Networks: Creating Decentralized Social Media on Top of Commodity Wireless Routers

Hiruncharoenvate, Smith, Edwards & Gilbert | GROUP 2016

supplement: system code

We present a new platform for building hyper-local social computing applications, running on home wireless routers via an underlying mesh network.

Understanding Anti-Vaccination Attitudes in Social Media

Mitra, Counts & Pennebaker | ICWSM 2016

What drives people to develop and perpetuate the anti-vaccination movement? Our results show that those with long-term anti-vaccination attitudes manifest conspiratorial thinking and mistrust in government.

Recovery Amid Pro-Anorexia: Analysis of Recovery in Social Media

Chancellor, Mitra & De Choudhury | CHI 2016

By developing a statistical framework using survival analysis we find that recovery on Tumblr is protracted - only half of the population is estimated to exhibit signs of recovery after four years.

#thyghgapp: Instagram Content Moderation and Lexical Variation in Pro-Eating Disorder Communities

Chancellor, Pater, Clear, Gilbert & De Choudhury | CSCW 2016

We present a quantitative study investigating pro-ED communities on Instagram in the aftermath of moderation and find that non-standard lexical variations of moderated tags are used to circumvent restrictions.

Algorithmically Bypassing Censorship on Sina Weibo with Nondeterministic Homophone Substitutions

Hiruncharoenvate, Lin & Gilbert | ICWSM 2015

Here, we show that it is possible to computationally generate homophone substitutions for banned terms on Sina Weibo, a technique that is difficult for the censorship apparatus to defend against.

Why We Filter Our Photos and How It Impacts Engagement

Bakhshi, Shamma, Kennedy & Gilbert | ICWSM 2015

We present a large-scale data analysis and in-depth interviews to understand filter-work. We find many use cases for filters, and that filtered photos are much more likely to be viewed and commented on.

CREDBANK: A Large-scale Social Media Corpus With Associated Credibility Annotations

Tanushree Mitra & Eric Gilbert | ICWSM 2015

supplement: data set

In this paper we present CREDBANK, a corpus of tweets, topics, events and associated human credibility judgements based on the real-time tracking of events on Twitter.

Open Book: A Socially-inspired Cloaking Technique that Uses
Lexical Abstraction to Transform Messages

Eric Gilbert | CHI 2015

best paper honorable mention

We introduce a technique called Open Book designed to address encryption's social usability problems. It uses data mining and NLP to make messages vaguer than the originals.

Comparing Person– and Process-centric Strategies
for Obtaining Quality Data on Amazon Mechanical Turk

Tanushree Mitra, C.J. Hutto & Gilbert | CHI 2015

best paper honorable mention

We 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.

In-group Questions and Out-group Answers:
Crowdsourcing Daily Living Advice for Individuals with Autism

Hong, Gilbert, Abowd & Arriaga | CHI 2015

We propose and evaluate a crowdsourcing approach to better support people with autism by offering rapid, concise, and socially appropriate coping strategies without compromising emotional support.

Piggyback Prototyping: Using Existing, Large-Scale
Social Computing Systems To Prototype New Ones

Catherine Grevet & Eric Gilbert | CHI 2015

best paper honorable mention

We propose a 6-stage prototyping mechanism for designing new social computing systems on top of existing ones. This allows a focus on what people do on a system rather than how to attract people to it.

Red, Purple and Pink: The Colors of Diffusion on Pinterest

Saeideh Bakshi & Eric Gilbert | PLOS One 2015

press: The Atlantic

We investigate whether there is link between color and diffusion. We find that color significantly impacts the diffusion of images and adoption of content on image sharing communities such as Pinterest.

Modeling Factuality Judgments in Social Media Text

Soni, Mitra, Gilbert & Eisenstein | ACL 2014

We obtain annotations of perceived certainty of quoted statements in Twitter. We find that readers are influenced by linguistic framing devices and do not consider other factors, e.g. sources, journalist.

Computing and Building Around Tie Strength in Social Media

Eric Gilbert | Foundations & Trends in HCI

This work presents a long-arc view of inferring tie strength via social media traces, and how we can alter interfaces to take advantage of it. Part of Eric's dissertation work, and set in the Twitter of 2010.

VADER: A Parsimonious Rule-based Model for Sentiment
Analysis of Social Media Text

CJ Hutto & Eric Gilbert | ICWSM 2014

supplement: data set code (github) python package

We present VADER, a simple rule-based model for general sentiment analysis, and compare its effectiveness to eleven typical state-of-practice benchmarks. We see it as a bigger and badder LIWC.

Faces Engage Us: Photos with Faces Attract More Likes and
Comments on Instagram

Saeideh Bakhshi, David A. Shamma & Eric Gilbert | CHI 2014

press: Mashable

This work finds that photos with human faces are 38% more likely to receive likes and 32% more likely to receive comments on Instagram, regardless of age and gender of the faces.

What If We Ask A Different Question?: Social Inferences Create
Product Ratings Faster

Eric Gilbert | CHI 2014

This paper studies eliciting product reviews as social inferences (i.e., "How do you think other people will rate this product?"). I find that they substantially reduce variance.

Overload is Overloaded: Email in the Age of Gmail

Grevet, Choi, Kumar & Gilbert | CHI 2014

We find that email overload, both in terms of volume and of status, is still a problem today. While work email tends to be status overloaded, personal email is also type overloaded.

Tensions in Scaling-up Community Social Media: A Multi-
Neighborhood Study of Nextdoor

Masden, Grevet, Grinter, Gilbert & Edwards | CHI 2014

This is a study of Nextdoor, a social media system designed to support neighborhoods. Nextdoor raises tensions in how it defines boundary of neighborhoods, and in the privacy issues it raises among its users.

Demographics, Weather and Online Participation: A Study of Restaurant Reviews

Saeideh Bakhshi, Partha Kanuparthy & Eric Gilbert | WWW 2014

supplement: data set | press: New York Times

This paper studies the effects of demographics and weather on restaurant reviews: restaurants in educated neighborhoods are highly reviewed and reviews written in nice weather are highly rated.

Managing Political Differences in Social Media

Catherine Grevet, Loren Terveen & Eric Gilbert | CSCW 2014

press: NBC Ars Technica

We investigate political disagreements on Facebook to explore the conditions under which diverse opinions can coexist online, and suggest design opportunities to bridge across difference.

The Language that Gets People to Give: Phrases that
Predict Success on Kickstarter

Tanushree Mitra & Eric Gilbert | CSCW 2014

press: Mashable Washington Post Forbes Fast Company

We explore the factors which lead to funding on Kickstarter. The language used in the project has surprising predictive power–accounting for 58.56% of the variance around successful funding.

Specialization, Homophily, and Gender in a Social Curation Site: Findings from Pinterest

Chang, Kumar, Gilbert & Terveen | CSCW 2014

We study two fundamental issues for social curation sites: flow of information & activities that attract followers. For example, sharing diverse content increases your following, but only up to a certain point.

Pair Research: Matching People for Collaboration,
Learning, and Productivity

Miller, Zhang, Gilbert & Gerber | CSCW 2014

To increase productivity, informal learning, and collaborations within and across research groups, we have been experimenting with a new kind of interaction that we call pair research.

A Statistical Framework for Streaming Graph Analysis

Fairbanks, Ediger, McColl, Bader & Gilbert | ASONAM 2013

We present a novel approach to the analysis of temporally varying networks that leverages time series and statistical techniques to quantitatively describe a social network.

Political Blend: An Application Designed to Bring People
Together Based on Political Differences

Doris-Down, Versee & Gilbert | C&T 2013

awarded best paper

In this paper, we introduce a mobile application called Political Blend designed to combat echo chambers: it brings people with different political beliefs together for a cup of coffee.

Analyzing Gossip in Workplace Email

Tanushree Mitra & Eric Gilbert | ACM Newsletter Winter 2013

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

"I Need to Try This!": A Statistical Overview of Pinterest

Gilbert, Bakhshi, Chang & Terveen | CHI 2013

supplement: data set

We use a quantitative approach to study three research questions about Pinterest. What drives activity? What role does gender play? Finally, what distinguishes Pinterest from Twitter?

A Longitudinal Study of Follow Predictors on Twitter

C.J. Hutto, Sarita Yardi & Eric Gilbert | CHI 2013

press: Huffington Post NBC Forbes

In this paper, we compare variables related to social behavior, message content, and network structure in order to interpret their relative impact to follower growth from different theoretical views.

Widespread Underprovision on Reddit

Eric Gilbert | CSCW 2013

press: Nieman Journalism Lab

In this paper, we present findings suggesting that widespread underprovision of votes is happening on Reddit. Notably, we find that 52% of the most popular links went overlooked on their first submisison.

Social Software as Social Science

Eric Gilbert | Digital Confidential (forthcoming, MIT Press)

In this chapter, we explore building social software to answer social science questions, covering issues like getting users and responding to the demands of the internet.

Have You Heard?: How Gossip Flows Through Workplace Email

Tanushree Mitra & Eric Gilbert | ICWSM 2012

press: Huffington Post Telegraph

Gossip is fundamental to social life. Here, we present the first large-scale study of it in cmc, looking at email where someone is mentioned in the message body but not included on the recipient list.

Designing Social Translucence Over Social Networks

Eric Gilbert | CHI 2012

best paper honorable mention

Social translucence is a landmark theory in social computing. However, we argue that it breaks down over modern social network sites and build a theory relating network structure to design.

Phrases That Signal Workplace Hierarchy

Eric Gilbert | CSCW 2012

press: All Things Considered

Hierarchy fundamentally shapes how we act at work. In this paper, we explore the relationship between the words people write in workplace email and the rank of the email's recipient.

Predicting Tie Strength in a New Medium

Eric Gilbert | CSCW 2012

best paper honorable mention

The term tie strength denotes the differential closeness with the people in our lives. In this paper, we explore how well a tie strength model developed for one social medium adapts to another.

Computing Tie Strength

Eric Gilbert | UIUC PhD dissertation, 2010

press: Technology Review

Relationships make social media social. But, not all relationships are created equal. This dissertation addresses this problem, merging the theories behind tie strength with the data from social media.

Widespread Worry and the Stock Market

Eric Gilbert & Karrie Karahalios | ICWSM 2010

press: Smart Money New Scientist

Our emotional state influences our choices. Here, we demonstrate that estimating emotions from blogs provides new information about future stock market prices.

Understanding Deja Reviewers

Eric Gilbert & Karrie Karahalios | CSCW 2010

awarded best paper

People who review products on the web invest considerable time and energy in them. So why would someone write a review that restates earlier reviews? We look to answer this question.

Predicting Tie Strength With Social Media

Eric Gilbert & Karrie Karahalios | CHI 2009

awarded best paper

Social media treats all users the same: trusted friend or total stranger, with little or nothing in between. In this paper, we present a predictive model that maps social media data to tie strength.

Blogs Are Echo Chambers: Blogs Are Echo Chambers

Eric Gilbert, Tony Bergstrom & Karrie Karahalios | HICSS 2009

Many commentators and researchers speculate that blogs isolate readers in echo chambers, cutting them off from dissenting opinions. Our empirical paper tests this hypothesis.

The Network in the Garden: Designing Social Media for Rural Life

Eric Gilbert, Karrie Karahalios & Christian Sandvig | ABS 2009

We know little about how rural communities use modern technologies. Using social capital theory, we predict differences between rural and urban users and find strong evidence supporting our hypotheses.

The Network in the Garden

Eric Gilbert, Karrie Karahalios & Christian Sandvig | CHI 2008

awarded best paper

We know little about how rural communities use modern technologies. To address this gap, we explore behavioral differences between more than 3,000 rural and urban social media users.

Using Social Visualization to Motivate Social Production

Eric Gilbert & Karrie Karahalios | IEEE MM 2008

In this article we argue that social visualization can motivate contributors to social production projects, such as Wikipedia and open source development.

A Social Visualization of Distributed Software Development

Eric Gilbert & Karrie Karahalios | Interact 2007

We present CodeSaw, a social visualization of distributed software development. It visualizes a distributed software community from two perspectives: code repositories and project communication.

The QuarkNet/Grid Collaborative Learning e-Lab

Marge Bardeen, Eric Gilbert, et al. | Journal of Grid Computing 2005

We describe a case study that uses grid computing techniques to support the collaborative learning of high school students investigating cosmic rays.