TED is an awesome platform for ideas, so we thought an interesting experiment would be having our API provide recommendations based on the talks’ transcripts.

Check it out - https://lateral.io/visualiser/ted

As our concept matching API does not use tags to make recommendations, it is interesting to see the suggestions it makes based on its conceptual understanding. We aim to source unexpected connections using this approach.

The results were captivating as we found ourselves conceptually jumping around TED talks, rediscovering old classics and uncovering interesting new talks.

The path you take is inherently tied to your interests. You begin by choosing a random talk or searching for one you already know. You then get a list of related talks, which you can select and again get related talks for. Hence, you always decide what step to take.

Next we will be mashing up different data sources, for example looking at mixing different NGO reports with TED talks to try to source solutions.

We’ll be writing up a post on how you can build recommenders for any mix of your favourite sites’ content very soon!