Keyword/phrase extraction from free text using NLTK and Python for structured queries

Question

I want to interpret specific keywords from a free text such as "I want to order boiled eggs and spinach soup from nearest restaurants' and use them to search content from my database. For eg i want to extract following specific keyword(s) and want an output similar to following like a tuple of (Keyword type , Keyword value). eg ('Food Item','Boiled Eggs') , ('Food Item','Spinach Soup') , ('Location','Nearest Restaurants') etc.

Need to use these values and types to further refine and to query my tables in database.

I was trying hard to find answer using various techniques using python and NLTK but need help to point me to right direction. Am i using the right techniques / frameworks? Need help here

Thanks, Sammy

Answer 1

It's not that difficult, but not that super precise either.

Use NLTK's PoS Tagging , then extract nouns and compound nouns , and use them in your DB queries.

A natural sentence is practically has two parts, an action, ie the verb, and a noun, ie the object.

Hence, "I want to order boiled eggs and spinach soup from nearest restaurants" tells us that:

• the action/verb is order
• objects/nouns are boiled eggs and spinach soup
• location is nearest restaurants