Gregg B. Johnson
Arts and Sciences
Sentiment Analysis on New York Times Coverage Data
Departmental Affiliation: Data Science/ Political Science
College of Arts and Sciences
The extant political science literature examines media coverage of immigration and assesses the effect of that coverage on partisanship in the United States. Immigration is believed to be a unique factor that induces large- scale changes in partisanship based on race and ethnicity. The negative tone of media coverage pushes non-Latino Whites into the Republican Party, while Latinos trend toward the Democratic Party. The aim for this project is to look at New York time data in order to identify how much immigration is covered in newspaper outlets, specifically Latino immigration, and to determine the overall tone of these stories.
In this research, we seek to determine individual articles take a positive, neutral or negative stance. We achieve this using a dictionary-based approach, meaning we look at individual words to assess if it has a positive, neutral or negative connotation. We train our data using publicly accessible sentiment dictionaries such as VADER (Valence Aware Dictionary and Sentiment Reasoner). However, this task can be difficult because certain words can be dynamic and may pertain to a positive or negative sentiment in context of the article. In order to resolve this issue, we use reliability measures to ensure that the words of high frequencies are in the correct sphere of negative, neutral, and positive light.
Information about the Author(s):
Faculty Sponsor(s): Professor Gregg B. Johnson and Professor Karl Schmitt
Student Contact: Gabriel Carvajal – email@example.com
Carvajal, Gabriel; Schmitt, Karl; and Johnson, Gregg B., "Sentiment Analysis on New York Times Articles Data" (2020). Symposium on Undergraduate Research and Creative Expression (SOURCE). 917.
For Artistic Presentations