Investigating News Source Characterizations Using Reddit Audience-based Metrics (undergraduate honor thesis)

Overview: In this thesis, I built embeddings to represent over 2000 online news sources using their sharing statistics on the social platform Reddit. I then evaluated the usefulness of this embedding system by trying to answer questions about the characteristics of those news sources and the relationships between them using clustering and classification methods. This is an interesting problem because a meaningful embedding can help us gain understanding about news sources based on how readers perceive and share them rather than relying on their self-reported characteristics. My thesis aims to do this in a general and scalable way since my method does not rely on Natural Language Processing (NLP) thus it is language agnostic, and instead rely on the widely used discussion and sharing site Reddit.

Tech used: Python sklearn for t-SNE, PCA, SVM, various clustering methods.

Find it here: Wellesley College thesis repository for my complete thesis.

Analysis of 2020 Election Media Coverage

Overview: I create and maintain a database containing data collected by the CredLab at Wellesley College which are the news articles included in the Top Stories panel with the name of those running for president in 2020 election as the query. I then created a responsive website which hosts interactive graphs visualizing the data. The goal of this project is to aid people (especially who are not computer scientists) to make sense of the data we collected to see how Google's algorithm works in deciding which news articles to highlight and to see whether political bias exists.

Tech used: Python, Flask, Plotly, MySQL, HTML, CSS, JavaScript

Wellesley in Time

Overview: Wellesley in Time is a geolocation-based Android application which uses the user’s location on Wellesley campus to display photographs of historical events and traditions that happened specifically around the user's current location, and provides reflection prompts for each photograph. If used while off-campus, then photographs are shown randomly.

Tech used: MIT App Inventor

Find it here: This link or search for 'Wellesley in Time' on Google Play Store.

Cov n Good News (web app)

Overview: Cov n Good News is a web application that allows users to see the current statistics of COVID-19 in the world as well as in individual countries, as well as provides a list of good news regarding COVID-19. The statistics and the list of good news are displayed side-by-side on the page as to show users that no matter how the COVID situation may look like, there are still good happening in the world.

Tech used: Python, Flask, AJAX (JavaScript), HTML, CSS, COVID-19 API, Google Custom Search API.