Automating Big Cat News: My Journey with NewsCatcherAPI and OpenAI
Since February 2007, I’ve been manually searching, sometimes translating, and publishing news about big cats — a journey that has spanned over 18 years. This process has always been time-consuming, involving extensive manual searching and processing of news articles. For years, I had wished to automate this task, and in the past month, I finally managed to do it in my free time.
I’ve implemented a news search using the API provided by @NewsCatcherAPI, a fantastic service with numerous news sources. I started by searching for Ukrainian news and found that this service has more Ukrainian sources than other similar APIs. While the number of Ukrainian sources is less than you might find on Google News, it’s still impressive.
@NewsCatcherAPI is free for non-commercial use, which suits my project perfectly since it’s non-commercial. The goal of my project is to popularize and provide fresh information to Ukrainians about big cats and to promote the Ukrainian language. The language is currently under threat from various forces and needs support. Besides, it’s simply pleasant to use my native language.
After retrieving the news, the results are filtered using basic algorithms. Then, neural network models GPT-4o-mini and GPT-4o analyze them to detect which wild cat species each article is about. Valid news articles are further analyzed by GPT-4o-mini to be tagged by the country they’re related to, and if it’s an event in Ukraine, the specific region.
A neural network ChatGPT-4o then translates news from English into Ukrainian. These translations are analyzed by ChatGPT-4o and o1 and edited by GPT-4o-mini multiple times to remove any signs of being translated, aiming to make them look and feel like articles written by native Ukrainian speakers. This level of automation is mainly possible thanks to the advanced OpenAI’s o1 model — it’s not a cheap solution, and I’m currently looking for alternatives. I prefer content that is easy to read with minimal literal translation.
After all this, I approve the news for publication, and they are posted on social networks and my website. I was able to create a new website from scratch in a couple of weeks. Laravel and @FilamentPHP were a great help. I could have developed the website faster, but my skills had become a bit rusty since I’ve been working more on parts of a complex web app than on complete website creation.
In my upcoming posts, I’ll try to delve deeper into how I implemented all this. The project is open source, and you can check it out on my GitHub. However, I must warn you that the code isn’t perfect in some places. The resulting website is located at https://bigcats.org.ua, and you can find social media links in the footer there.
