Week 9: Presentation And Future Applications

This week, I began preparing for my final presentation and figuring out the applications of my senior project.

For the first half of my week, I worked on the slideshow for my senior project. First, I began by explaining the effects of fake news. Fake news has pervaded almost every corner of the internet, from fake articles on the web to falsified information on social media. I also broadly defined fake news as inaccurate or misleading content posed as legitimate news, often pervasive and inciting in nature. Additionally, I named a few harmful groups such as QAnon, InfoWars, and American Independent, which have spread hatred and fabricated propaganda targeted at specific groups of people that have difficulty distinguishing between real news and fake news. Ultimately, I reminded the audience that fake news can lead to serious consequences, such as loss of confidence in legitimate media organizations.

I also gave some real examples of fake news that harmed people in real life. One example I gave was Pizza Gate, a scandal on Twitter in which some Trump supporters and Twitter bots began spreading a fake story about Democratic leaders that created an underground pedophilia ring in the basement of a pizzeria called the Comet. This story went viral on Twitter, with over a million messages including the word “Pizzagate” circulating within a month. This led to a shooting, in which a man open-fired into The Comet with the intention of liberating the children who were allegedly being held in the basement.

At this point in my presentation, I began to describe my proposed solution. The immense volume of false information is a problem that can be mitigated with machine learning. If algorithms could check the validity of articles on the internet and flag them for false content, this would significantly lower the risk of deceiving information influencing users. This brought me to the purpose of my project, which was to explore more NLP techniques in an effort to increase efficiency. Thus, I explained that I would be constructing the classic RoBERTa model and tracking its accuracy using F1-Score, then creating a custom-LSTM model from scratch, a different type of model that I believed to have better capabilities for text classification.

Afterward, I made sure to explain what NLP was, what an LSTM was, and the classic RoBERTa model. I also explained how RNNs and LSTMs worked and why they were very similar models. Next, I explained the process of choosing a dataset, preprocessing data, coding the model, and testing it using Wandb. I also touched upon the challenges I encountered during my project and the future applications of my research.

I began working on the future applications of my research in the second half of the week, which included creating a Chrome extension (or iPhone app) that could possibly utilize my work to flag content online as potential misinformation. This last part of my project is still underway and is something I can work on even beyond my senior project. I am excited to keep working and see where it takes me!