News-Sentiment-Analysis (CSI-280-51 Project)

By Kimberly Benson, Thomas Lukasiewicz, Griffin Gooch-Breault, Yin Bo Chen

What Is This Project?

The goal of this project is to allow for a user to feed a pre-trained AI a news headline to pick out key words and sentinment (positive, neutral, negative) within it. By doing this, users can understand how a news headline skews in terms of sentiment, and compare headlines on the same subject to other news outlets.

Repository Layout

ASBA

This directory is the LLM model and its files.

ABSA.py - Python file that handles the LLM's logic. requirements.txt - set up for LLM

We are using a model devleoped off of microsoft's deberta V3.

Please view the papers of its creators here:

Yang, Heng, et al. “Back to Reality: Leveraging Pattern-Driven Modeling to Enable Affordable Sentiment Dependency Learning.” CoRR, vol. abs/2110.08604, 2021, https://arxiv.org/abs/2110.08604.

Yang, Heng, et al. “PyABSA: A Modularized Framework for Reproducible Aspect-Based Sentiment Analysis.” Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, CIKM 2023, Birmingham, United Kingdom, October 21-25, 2023, edited by Ingo Frommholz et al., ACM, 2023, pp. 5117–22, https://doi.org/10.1145/3583780.3614752.

Flask

This directory handles the website for the project. Static subdirectory handles the css and json, templates contains the html.

app.py - File that runs the website and handles the main logic.

dataset

This directory contains various CSVs of real news headlines from Jan - March 2025. python_files contains the .py files used to collect data, and raw_news_titles contains the CSVs.