Week 1: Literature Review

Hi everyone! I’m Harish, and I am excited to share my first blog about my research.

This project focuses on understanding echo chambers on social media, and how we can measure and detect them using data. Specifically, I am studying platforms like Reddit and X to see how conversations cluster and why certain communities become ideologically isolated. 

Background and Introduction

I became interested in echo chambers after noticing how online discussions about the same event can feel completely different depending on where the discussion happens. In some spaces, everyone seems to agree on the same idea, and in others, the narrative can be entirely different with discussions being very diverse. 

An echo chamber is an environment where people are mostly exposed to opinions that reinforce their own existing beliefs. Instead of seeing a wide variety of opinions and perspectives, users mainly talk with others who think similarly, which can reinforce viewpoints and lead to minimal disagreement. This way, a person’s ideas are “echoed” back to themselves. 

Because so much of public conversation now happens online, understanding these environments is important for studying polarization and information spread.

How I am studying echo chambers

My research combines natural language procession with network analysis to detect echo chambers in a measurable way. Rather than relying on anecdotes, I am building a framework that looks at both what they say and how they interact.

There are three main parts that I am focusing on:

Network Structure looks at who interacts with whom. If users mainly reply to or share similar content from the same group of people, it shows a tight and connected cluster with limited outside interactions.

Topic homogeneity is a metric to see how similar conversations are within a community. When most discussions revolve around the same topic or idea, the group shows strong alignment to what they are discussing.

Sentiment homogeneity captures whether people express similar feelings toward a topic. When a community consistently reacts in the same positive or negative way, it reinforces shared perspectives and can bring the community closer together.

What I have done so far

This week, I focused on conducting a literature review and thinking about how my model will be built. I read research papers on echo chambers and polarization to figure out the importance of the three components I mentioned previously. My goal was to see how other researchers measure different network metrics like isolation, homophily and content similarity that I will incorporate into my own project.

As I looked through these papers, I mainly looked through the methodology sections to see what metrics were used and why. For example, I looked at different ways to measure how separated communities are in a network, as well as different techniques to see how similar conversations are in a group. 

I also began working on the main product of my project: deciding the weights I will be using on my components of the Echo Chamber Index, which is a quantitative sum of the three parts of an echo chamber mentioned before which are network structure, topic homogeneity, and sentiment homogeneity. Since each of these parts may influence an echo chamber differently, I want to add weights to each part to make my index value more accurate. As I begin working on my index in the future, I will most likely begin with equal weights on all three components and adjusting them based on testing. 

To test these weights, I will be using sensitivity analysis. This means I will make slight changes to the weights and see how the final value changes. If the output changes a lot, I will change the weights and may make changes to the components themselves. 

Why this matters

By measuring these three factors together, I hope to build a clearer picture of how echo chambers form and how strong they are. This research can help us better understand polarization online and bring us one step closer to bringing diversity to these communities. 

In future posts, I’ll share more about my methodology, the data I will be using, and what I discover as I conduct this research. 

Thanks for reading, and I’m excited to share more as the project grows!