Natural language processing, how right?

Question

Python

Let's say there is a phrase "Reserve a place in a cafe." How to parse it correctly?
There is an idea to use Syntaxnet from Tensorflow. He knows how to determine dependencies and parts of speech (then it should be easy) and multilingual + support from Google. But there are problems with its installation and it works as a separate script, and not as a library in python, which is extremely inconvenient.
I saw in this lecture Tensorflow can parse search queries, but I did not understand how to do this. Maybe there are analogues or who knows how to do it like in a lecture?

Answer 1

1. You need to learn how to determine the part of speech of all words (+ case, number, gender, etc.) in a sentence using ready-made libraries.
2. Build a "tree" of relations (dependencies) of words (parts of speech) using complete templates of prepared "chains" of words that may be present in the sentence (create manually or train the NN vector).
3. Have a dictionary of imperative moods and algorithms for the necessary actions for them.

Answer 2

NLTK?

Answer 3

Search queries will be quite difficult to parse, because more often than not, a request is not a complete sentence.