Why not both?
I use traditional scrapers to get the data, then LLMs to clean/parse it. Best of both worlds.
For example, scrape with Selenium, then dump the messy output into GPT-4 to structure it. Works way better than relying on the LLM alone.
I use traditional scrapers to get the data, then LLMs to clean/parse it. Best of both worlds.
For example, scrape with Selenium, then dump the messy output into GPT-4 to structure it. Works way better than relying on the LLM alone.
