Project Philosophy

Navigating the Information-Polluted World.

Two men in lab coats.

Our information supply is contaminated. In an era where generating and publishing content is remarkably easy, we are bombarded with irrelevant, redundant, unsolicited, incorrect, and otherwise low-value information.

We, as researchers in Natural Language Understanding, developed this site to facilitate the research on computational methods to address these issues. Specifically we are interested in developing machine learning and inference methods for understanding documents along the aforementioned dimensions.

Two men in lab coats.Man in dark room with multiple led lights.
Two men in lab coats.Man in dark room with multiple led lights.City view from above.
519E1E72-B8F6-4980-843F-CA239029F2CA 519E1E72-B8F6-4980-843F-CA239029F2CA 519E1E72-B8F6-4980-843F-CA239029F2CA
Our Vision for the Site

Case Study: COVID-19

Define & Address the Research Challenges

The spread of COVID-19 also gives rise to the proliferation of information pollution. We would like to view this as a case study for us to define and address the key research questions raised by the need to navigate our way through the information-polluted space

In the context of COVID-19, we would like to provide a reasonably-sized set of relevant, trustworthy, and timely information that allows users to become quickly informed about specific topics of interest.

Challenges

To reach our goal of using computational methods to select read-worthy information, we need to overcome many challenges. Here are three main objectives we would like to highlight: (1) Understand Perspectives in Documents, (2) Identify Trustworthy Sources, and (3) Reduce Information Redundancy. We hope to build upon current information recommendation systems and organization schema, and take one step closer to the three objectives above. We hope that our platform will allow critical readers to become more informed about COVID-19.

Call For Help

Our site and underlying machine learning models are open-source. We invite researchers in related fields to join the effort on the project, and potentially contribute to our current or future research efforts.

01

Understand Perspectives in Documents

Documents on similar topics could present different perspectives, and so have different implications. We think it is important to present information in a way that such nuances are visible to readers.

02

Identify Trustworthy Sources

Trustworthiness of the information sources is often a good indication of the reliability of the information itself.

03

Reduce Information Redundancy

To overcome the trade-off between the selection-bias versus the limited amount of information we are able to present, we seek to present information with a spectrum of diverse perspectives.

Meet the Team

Who are we?

Dan Roth

Dan Roth

Professor of Computer and Information Science

Sihao Chen

Sihao Chen

Computer and Information Science Ph.D. student

Xander Uyttendaele

Xander Uyttendaele

Research Assistant