I’m currently working on building a service that extracts keywords from articles. I’ve crawled approximately 700 articles. However, I need help creating the dataset, so I’m writing this post. Which of the following three options should I pursue?
-
List the articles to be input into GPT, and have humans manually extract keywords from them to create a dataset in CSV format…
-
List the articles to be input into GPT and use GPT to extract keywords to create the dataset.
-
Or should I directly apply the list to the GPT model, using it as a knowledge base, without any data preprocessing?
I’m still learning about AI , and I’d like to discuss this matter as I’m inexperienced. 
