Kumu meet Bhindi (bh – that’s an aspirated version of b)
Interactive network visualization of bhindi using Kumu.io
In predominantly vegetarian India, bhindi is generally well liked. Bhindi is pronounced as bin-dee (with bh as a single consonant instead of ‘b’; see this guide for a better description on pronounciation – search for ‘bh’). Bhindi is also known as okra or ladies finger. I think in a dinner party I had overheard someone say, ‘Wow bhindi. I love bhindi’ (could have been any of my friends Rajeev, Aditya, Deepti, Sumana, Raghu, Advait, Pavan or Sriram).
In this visualization I have shown instances of ingredients when they appear together in a recipe. When ingredients appear multiple times, it might mean that they taste well together. When they don’t appear together it might mean they clash.
I used recipes from harekrsna.com, since they are traditional, home cooking type recipes. The recipes are also from different regions in India. Click on the image below to access the visualization on Kumu. Try hovering over the nodes and connections. Click on a node to see the related recipes on the side panel.
Here is an example of food pairing of fenugreek/methi seeds. In the recipes I have chosen fenugreek appears in around 8 recipes. It is used in many baingan/eggplant recipes and not so much with bhindi.
These are some questions you could ask:
- curry leaves has a strong link to black mustard seeds, but not to nigella/kala jeera
- amchur doesn’t seem to gel with fenugreek/methi or nigella/kala jeera
- coriander seeds pair with eggplant, but not with bhindi
Keep mind that I have done a very small sample which is just a teaser. I again recommend viewing Y Y Ahn’s video to understand this topic.
Here is a quick guide on how to use the tools on Kumu.io:
- Hover over an element to see the connected elements
- Click on the element to see the recipes in which it is used in the left Pane
- Click on a connection to see which recipes both ingredients are used
How did I create the data
I parsed the html using Ruby/Nokogiri to extract the ingredients for each recipe. I started by writing some code to remove the measurements and qualifying words, but it turned out to be too much work. It took me around 20 minutes to manually create a list of ingredients.