Scroll to begin
Hello, I'm Naavya
Where it started
Merging art with code.
Where it went
to create unique data art.
Project 01
Generative Data Art
The paper
Network Analysis and Generation of Data Art from Restaurants Data
This all grew out of a co-authored paper on Bangalore's restaurant scene. We pulled Zomato data - cuisines, menus, featured dishes, reviews - and built a similarity network across 8,792 restaurants: TF-IDF embeddings per place, then nearest-neighbour matching on the twenty closest, quantile-weighted into 147,424 edges.
Community detection ran four ways - Label Propagation, Louvain with a Constant Potts Model, Demon, and Bayesian statistical inference. Demon read best: it grouped restaurants by region - Tamil Nadu, Bihar, the Delhi-Punjab-Haryana belt - with no location data in the graph at all.
A bar chart would have told you the degree distribution follows a power law. It would not have shown which communities sit next to each other, or which restaurant is quietly bridging two cuisines it has no business bridging. So we turned the analysis into graphs.
Read the paper (PDF)- Restaurants
- 8,792
- Unique edges
- 147,424
- Average degree
- 33.54
- Network diameter
- 8
- Communities (Demon)
- 177
- Built with
- Python, Gephi, Processing
The graphs
Look at the graphs below.
Same network, different layouts. Colour marks community; size follows centrality. Click the stack to flip through.
Project 02
Generative Runs
Used Java to run a generative algorithm in a way in which no output is ever the same - it builds unique patterns by picking different hexadecimal codes through a random algorithm.
Project 03
AI Image Dashboard
Microsoft Ethical Tech Summit
Worked with Professor Rhodes to create an interactive AI dashboard for the Microsoft Ethical Tech Summit.
We generated images of people from different parts of the world telling their stories through AI generated avatars and voices. An attendee at the tech summit could step close to the dashboard, and interact with any avatar, almost making it life like.