Date of Graduation

5-2025

Document Type

Thesis

Degree Name

Bachelor of Science in Computer Science

Degree Level

Undergraduate

Department

Computer Science and Computer Engineering

Advisor/Mentor

Gauch, Susan

Committee Member

Zhang, Lu

Second Committee Member

Yang, Song

Abstract

Understanding public sentiment on social issues is crucial for gauging the stance of the general population. Traditionally, surveys have been a common approach for this. However, to capture more candid opinions, social media provides a rich source of unadulterated opinions. By analyzing social media statements, we can gain insights into the perspectives of specific groups. More specifically, we will investigate the attitudes of the public into the relationship between hard work and success in the workplace.

Using a neural network trained on tweets from X, formerly known as Twitter, we can categorize each tweet as either pro-luck or pro-meritocracy. But will this neural network, trained on a particular set of tweets, accurately assess sentiment on slightly different topics from the same platform? This is where the concept of transfer learning comes into play. We have trained our model with certain topics, but we are interested in testing it against related, but different topics. Transfer learning is a useful technique in neural networks since it eliminates the need to train a neural network from the start, which could be both time consuming and costly. However, it's crucial to evaluate the effectiveness of the transfer learning model to gauge its reliability.

Keywords

Sociology of Work and Success; Computational Social Science; Digital Sociology; Neural Networks; Data-Driven Social Research; Transfer Learning

Download

Share

COinS