Mission

The Socially Embedded Lab at Ben-Gurion University studies social behavior in large-scale, online information systems. The lab investigates areas where information systems are suboptimal for people -- for example, by not meeting people's needs, goals or expectations -- and proposes new computational measures to bridge the gaps. For example, it studies the scale and scope of fake news on social media and develops methods for limiting its spread, audits large-scale algorithmic systems such as Google Search and Gmail to reduce bias and inequality, and examines online user engagement with news in order to promote better measures for quantifying it. The lab is part of the Software and Information Systems Engineering (SISE) department at BGU.

News

Sep. 2023

Arnon Strum (BGU) and Nir Grinberg are awarded a 2-year grant from Israel's Innovation Authority to develop NLP resources and models for the Israeli Criminal Justice System.

July 2023

Nir Grinberg wins one of two Best Parallel Talks at the 9th International Conference on Computational Social Science.

Apr. 2023

Nir Grinberg receives BGU's Teaching Innovation Award for 2021-2022.

Dec. 2022

Talia Meital Schwartz (BGU) and Nir Grinberg are awarded a 3-year grant from Israel's Ministry of Science and Technology to study out-of-home placements of at-risk children in the Israeli social welfare system.

Nov. 2022

Jennifer Oser (BGU) wins an ERC Starting Grant with a Work Package co-led by Nir Grinberg.

People

Nir Grinberg

Assistant Professor

Sahar Baribi

M.Sc. student

Assaf Shamir

M.Sc. student

Moran Shay

M.Sc. student

Eitan Torf

M.Sc. student

Hagit Liven Bar

M.Sc. student

Boris Sobol

M.Sc. student

Miri Yitshaki

M.Sc. student

Zahi Mizrahi

M.Sc. student

Ori Peri

M.Sc. student

Din Ezra

M.Sc. student

Projects

Fake news on social media

Who is sharing and who is consuming fake news on social media? What is the impact of misinformation? How can we robustly detect misinformation before it becomes widespread? What actions can we take against online misinformation and what are their ramifications? A number of research projects in the lab explore these questions using a combination of quantitative and qualitative research methods.

Bias in AI-mediated communication

Human communication is no longer "just" mediated by computers. A growing number of interactions are now fully-generated by AI with light supervision by humans through the use natural language generation models and machine learning. A prime example of that is Google's Smart Reply, which offers Gmail users three short but complete reply suggests to save them time responding to emails. And while the time-saving nature of these technologies is clear, they also raise an array of questions about the short- and long-term biases that they introduce, both in the resulting communications and the relationships behind them. We study these questions using a mixed-methods apporach that includes algorithmic audits, surveys, and experiments.

Predicting unemployment using social media

Up-to-date information about individuals' employment status is of tremendous value for a wide range of economic decisions, from hiring decisions to government policy. Yet, most current estimates are based on representative surveys that are costly, slow to produce, limited granularity, and somewhat inaccurate. These issues are even more extreme in low- and middle-income countries, where surveying infrastructure is not as developed as in high-income countries. Can information from social media contribute and complement existing estimates of labor market flows? In this line of work, we are using novel NLP techniques to accurately identify public disclosures about unemployment and develop statistical models to aggregate this information robustly over time.

Exposure to politics on social media and its impact on political behavior and attitude

As social media expands as a vector for news, it is increasingly important to understand the interplay between exposure to politics on social media and offline political behavior and attitudes. Prior work examined several aspects of media exposure and political participation, both in the general population and for specific socio-demographic groups. This line of work extends prior work by characterizing the different types of political exposure on social media and examine the causal mechanisms that influence political behavior and attitude. To that end, we are using observational data analysis and seek to combine it with survey-experiments.

Searching for science all over the world

Search engines are one of the primary sources for people to learn about science. The quality of scientific information found there is critical for students to make their first steps in acquiring knowledge and skills in science, and for non-scientists to navigate everyday life decisions using science. Despite the growing availability of information on the Web and increased internet access, inequalities persists across different cultures and languages. Do search engine work to reinforce existing inequalities or close the gaps? In this work, we focus on the quality of scientific information available on Google Search all over the world and across basic scientific concepts (e.g. mass), pressing scientific issues (e.g. global warming), and popular science concepts (e.g. electric cars). We approach this problem using an algorithmic audit of search results and through human-assessed quality by experts and volunteers.

Engagement with news

Despite the importance of news, we know relatively little about how well journalism is achieving its goals of informing citizens and fostering healthy discussion in society. In the absence of good metrics for measuring these important goals, both systems and people in newsrooms resort to shallow and incomplete measures of user engagement such as clicks, likes, or total time spent on a page. The goal of this research is to take a closer look at the different ways in which people interact with online news, and proposes new measures to help both information systems and newsrooms gain better insights about readers and content. To that end, we combine in-depth analysis of post-click user behavior on news articles, modeling of article content, and knowledge obtained using lab experiments about reading to devise new measures for quantifying reading online.

Publications

Pandemics, Protests, and Publics: Demographic Activity and Engagement on Twitter in 2020, Sarah Shugars, Adina Gitomer, Stefan McCabe, Ryan J. Gallagher, Kenneth Joseph, Nir Grinberg, Larissa Doroshenko, Brooke Foucault Welles, and David Lazer. Journal of Quantitative Description: Digital Media 1, 2021.
pdf | code

The COVID States Project #18: Fake News on Twitter, David Lazer, Damian J. Ruck, Alexi Quintana, Sarah Shugars, Kenneth Joseph, Nir Grinberg, Ryan J. Gallagher, Luke Horgan, Adina Gitomer, Aleszu Bajak, Matthew A. Baum, Katherine Ognyanova, Hong Qu, William R. Hobbs, Stefan McCabe, and Jon Green. OSF Preprints 2021.
pdf | demo

Tackling misinformation: What researchers could do with social media data, Irene Pasquetto, Briony Swire-Thompson, Michelle A. Amazeen, Fabrício Benevenuto, Nadia M. Brashier, Robert M. Bond, Nir Grinberg et al. Harvard Kennedy School Misinformation Review 1, no. 8, 2020.
pdf

Fake News on Twitter During the 2016 U.S. Presidential Election, Nir Grinberg, Kenneth Joseph, Lisa Friedland, Briony Swire-Thompson, and David Lazer. Science, Vol 363, Issue 6425, pp. 374-378, 2019.
pdf | supplementary | data | code

Reports of the workshops held at the 2016 International AAAI Conference on Web and Social Media, Jisun An, Ciampaglia, Giovanni Luca, Nir Grinberg, Kenneth Joseph, Alexios Mantzarlis, Gregory Maus, Filippo Menczer, Nicholas Proferes, and Brooke Foucault Welles. AI Magazine 38, no. 4, pp 93-98, 2017.
pdf

Combating fake news: An agenda for research and action, David Lazer, Matt Baum, Nir Grinberg, Lisa Friedland, Kenneth Joseph, Will Hobbs, and Carolina Mattsson. Harvard Kennedy School, Shorenstein Center on Media, Politics and Public Policy 2, 2017.
pdf

Understanding Feedback Expectations on Facebook, Nir Grinberg, Shankar Kalyanaraman, Lada A. Adamic, and Mor Naaman. CSCW 2017.
pdf

Changes in Engagement Before and After Posting to Facebook, Nir Grinberg, P. Alex Dow, Lada A. Adamic, and Mor Naaman. CHI 2016. Honorable Mention
pdf

Contact

Still haven't found what you're looking for? Feel free to get in touch with any lab member for further questions and suggestions.

Looking to join the lab? Please contact Nir AFTER you've read some of our papers. Don't forget to attach your CV and transcripts, and make sure to specify at least one research direction that excites you and any special skills that set you apart.