In this report, I analyze individuals’ social media habits using Gephi for network visualization. My goal is to reveal patterns in users’ engagement across platforms like Tumblr and Twitter and identify similarities in the types of content shared. This analysis aims to uncover patterns in how different social media platforms are used together, providing insights into the interconnected nature of online social behavior.
Data Collection and Preparation
To collect data for this network visualization project, I conducted a poll using a Google Form due to the lack of publicly available datasets. The form included questions like “Which social media platforms have you ever used?” and “Which platform do you currently use?” to understand participants’ engagement across various platforms.
Additional questions like “Do you access social media via desktop or app?” and “Which platform do you use most frequently?” provided context on preferred access methods and dominant platforms. Hopefully, In the future, these insights will help me understand how information spreads across platforms, identify content trends, and predict shifts in user behavior.
After collecting 28 responses, I organized the data in a Google Sheet, isolating responses related to social media platform usage. To facilitate further analysis, I needed to transform the data format. As the responses were initially listed in a single column separated by commas, I utilized Excel’s text-to-columns function to split them into individual columns. This process effectively expanded the data set, resulting in a sheet with 15 columns representing each platform, with participant responses arranged in rows.
Next, I imported the cleaned data into R for further manipulation. Using R formulas, I calculated direct and indirect links between platforms and determined connection weights for network visualization. This process transformed raw survey responses into a structured dataset, suitable for generating insightful network visualizations using Gephi.
![](https://i0.wp.com/studentwork.prattsi.org/infovis/wp-content/uploads/sites/3/2024/05/Social-Network-Gephi-1024x500.png?resize=840%2C410&ssl=1)
Graphing for Insights
After importing the network data into Gephi, I refined the graph by adding labels to nodes and adjusting their size and color to effectively convey information. Additionally, I used the Yifan Hun layout to adjust the partition in nodes. Larger nodes indicated higher usage frequency, while colors distinguished platform types. I noticed Gephi’s automatic color blending for edges with equal weights, this caused me to seek a way to add a legend for clarity.
The insights that I gained from the analysis revealed substantial audience overlap among platforms like Facebook, Instagram, and Twitter, suggesting favoritism among users in sharing pictures and updates. Notably, LinkedIn exhibited significant overlap with Facebook and Twitter, most likely due to the three platforms being most popular among professional content sharing. Similarly, Snapchat and TikTok showed considerable overlap, driven by younger demographics. However, YouTube emerged as the most interconnected platform of all.
While the Gephi graph provided valuable insights, its complexity made it challenging to interpret due to the multitude of connections. To improve readability, I explored alternative visualization methods. This included creating an interactive radial network graph using Flourish. This approach offered several advantages, including the incorporation of brand logos alongside platform names, enhancing visual recognition. Additionally, the graph categorized social media platforms by color, facilitating easier identification and interpretation of groupings.
![](https://i0.wp.com/studentwork.prattsi.org/infovis/wp-content/uploads/sites/3/2024/05/Screenshot-2024-05-07-at-11.42.39 PM-1-836x1024.png?resize=836%2C1024&ssl=1)
User Feedback
After testing the network graph created in Gephi, users found it useful for exploring connections between social media platforms. They could click around and filter edges by node, simplifying their analysis. Incorporating user feedback, I aimed to enhance clarity by adding a legend to explain line colors and thicknesses. To achieve this, I took a screenshot of the Gephi graph and integrated it into a Tableau (click link for accurate depiction) dashboard.
Additionally, I developed a heatmap illustrating social media crossovers, complementing the network graph. This dual representation allows users to grasp relationships and data more intuitively. While the Gephi version offered interactive features such as hovering to filter each node’s edges, this functionality didn’t transfer seamlessly to Tableau. Ideally, future iterations would ensure that interactive elements remain intact when transitioning between platforms.
![](https://i0.wp.com/studentwork.prattsi.org/infovis/wp-content/uploads/sites/3/2024/05/Screenshot-2024-05-08-at-12.46.39 AM-1-1024x903.png?resize=840%2C741&ssl=1)
n conclusion, the visualizations have provided the insights I sought, with YouTube emerging as the most used platform, surprising me. Other findings have confirmed my initial hypotheses.
Moving forward, I see potential in further data collection and development to explore content sharing dynamics and misinformation spread. With continued refinement, these visualizations can offer deeper insights into online behavior and inform strategies for navigating digital landscapes effectively.