Story graph of 5000 TED Talks

Do you wonder how the TED talks narratives flow? Do you wonder if there are any similarities in terms of the ebbs and flows of emotions in each of these narratives.

Building upon my earlier project of open sourcing 5000+ TED Talk narratives, I did something fun. For each of these talks, I built a story graph. Essentially, it is a graph of how the emotions (sentiments, magnitude of the sentiments) vary over the narrative of the entire talk. The results, if I may so, are magical. 

Now you understand the reason for my short hiatus from the blog. I was building this.

Head out to https://labs.saranyan.com/projects/ted/storygraph and play with it. Let me know what you like and don't. One thing that I am still working on is similarity search based on the story telling pattern. The intent is to cluster similar speakers around audience engagement, story telling style, narratives and words used, etc. 

Here are the story graphs from some of my personal favorites.

Screen Shot 2018-05-06 at 4.15.46 PM.png

The art of stillness

Pico Iyer

Screen Shot 2018-05-06 at 4.26.11 PM.png

Bring on the learning

Sir Ken Robinson

Screen Shot 2018-05-06 at 4.29.00 PM.png

The price of shame

Monica Lewinskey

Screen Shot 2018-05-06 at 4.31.28 PM.png

The power of vulnerability

Brene Brown

Screen Shot 2018-05-06 at 4.33.01 PM.png

Your elusive creative genius

Elizabeth Gilbert

Screen Shot 2018-05-06 at 4.34.50 PM.png

The danger of a single story

Chimamanda Ngozi Adichie

Screen Shot 2018-05-06 at 4.48.29 PM.png

Haunting photos of polar ice

Camille Seaman

Screen Shot 2018-05-06 at 4.50.06 PM.png

The augmented reality of techno magic

Marco Tempest

Hope you like it. If you do, be an angel and spread the word. 

 

Exploring Harvard Business Review through AI

From 1956 - 2017, Harvard Business Review has published over 6500 articles (6656). And these articles cover 148 topics. Over the last few weeks I have been building bunch of tools to extract and analyze the data from HBR. In particular, I wanted to understand what are the topic trends over the years. 

Once I had extracted that bit, I noticed few interesting themes emerge across the spread of articles. I built an AI engine that extracted themes out of each of these articles. Here is the first iteration of this work.

This is a work in progress. For now, you can view the trends, click on a topic to view all the articles in that topic, and click on the themes (autogenerated by my AI engine) to explore articles further. I will add more stuff to this, but for now, enjoy this offering.