In May, I announced that I had completed building my website WordLink. I have recently decided to shut down WordLink, and this post briefly discusses that decision.

My primary aim for the project was to build a modern web app from scratch and to learn more about the different components that go into such a site. In this regard, the project was a success, introducing me to the Springboot framework, building CI/CD tooling, terraform and AWS, as well as much more. Please read my previous post for more information about the different parts of the project.

WordLink uses data from Wiktionary to find words with a similar etymology in different languages. It does this by creating a graph whose nodes are words and whose edges are the various etymological links between the words. The approach suffers from poor data quality because the etymological links are inconsistently recorded in Wiktionary.

There are two ways this might be addressed while keeping a similar approach:

  1. Invest a lot of engineering time into improving the parsing of Wiktionary data and potentially editing pages to make the links more consistent.
  2. Change the dictionary/etymology data source.

Alternatively, we could consider new ways to solve the problem. In particular, I recently found out that ChatGPT, and to a lesser extent Bard, can solve this problem quite well.

ChatGPT finding words related to "albero"

A new approach would be to formulate queries like the above and send them to the language model via some API. The language model’s output could be parsed to produce the similar words.

At this point, however, I want to invest my time into other projects. While I could leave the website as-is, I’m opting to close it completely to save on server costs.

The current best solution to finding etymologically similar words that I’m aware of is typing this phrase into ChatGPT.

List some English words that have similar etymology to the Italian word “albero”

Screenshot of the WordLink homepage

Screenshot of the WordLink search results for "jour"