ditrit / OGrEE_NLP

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

THE PROJECT

This project, sponsored by Orness, aims to transcribe instructions written in natural language into commands understandable for the CLI.
This plugin integrates with OGrEE, a digital twin tool.

The plugin is divided into 2 parts : the understanding of the sentence and the smart placement (solver). The understanding part can be autonomous.

See an example below :

NLP Processing example

CAUTION : This whole plugin uses the large model en_core_web_lg. It has large memory size : one could use the medium size model (replace lg with md). The small model will NOT work (see https://spacy.io/models/en).

Note : You must download the model on your computer / container : https://spacy.io/usage/models

IMPORTANT DETAILS

The library used to perform NLP processing is Spacy. Learn more here : https://spacy.io/usage/linguistic-features

This plugin is NOT AI-powered or ML-powered. It is using logical arguments to get comprehensive understanding of a sentence.
The use of spacy was precisely chosen to enable a future implementation of machine learning.

All the criteria used come from the wiki : https://github.com/ditrit/OGrEE-Core/wiki

INSTRUCTIONS

To launch, execute ogree_main.py

  • Please only give 1 instruction per sentence (creation, destruction or modification of 1 object)
  • Write explicitly the name of entities : "building", "rack"...
  • Be precise, a lack of information could lead to non understanding
  • Write names in upper case

About


Languages

Language:Python 93.4%Language:Jupyter Notebook 6.6%