I pushed parse_html.py to index folder. It's very short, but it's not working as I would expect in an ideal situation. I'm very stuck with parsing it properly because the side links. For instance, a chicken pot pie page would end up having muffin in it. I'm worried that indexing with such dirty file may result in bad search. Please feel free to change the code to make it better! Thanks :)

I'm struggling with search accuracy. Specifically, with the current tokenization framework, let's say a user search 'chicken', I don't think if makes sense to show any recipe that uses "chicken broth", instead of "chicken", and vice versa.

I've been reading some paper on phrase search, but haven't found anything particularly helpful.