Interactive Synthesis

This is a server that accepts a single example from a LTLTalk user together with the corresponding natural language definitions, and tries to generalize it to a definition.

Setting up the environment

create a virtual environment for python3 (for example, using virtualenvwrapper)
run pip install -r requirements.txt
run export FLASK_APP=flask-routes.py (or add it to the .bashrc file)

Running

flask run. (The server will run on the port 5000 by default)

Interacting with the whole system

Setup

Install frontend, semantic parser, and interactive synthesis engine locally. To install each part, please follow the instructions in the corresponding repositories.

Basic system functionality

In order to have a view of the overall system (as shown in Fig. 2 of the paper), do the following: Terminal 1:

navigate to the ltltalk-backend directory and run ./interactive/run @mode=ltltalk

Terminal 2:

navigate to interactive-synthesis directory
run `workon ltlTalk' (to use the correct python virtual environment)
run `flask run'

Terminal 3:

navigate to the frontend directory
enter folder voxelurn
run yarn start

Terminal 1 runs the naturalization server (described in Section 5 of the paper). Terminal 2 runs the interactive synthesis server (described in Section 4 of the paper). Finally, Terminal 3 runs the frontend. Upon running yarn start, a firefox window should open. If this does not happen, visit localhost:3000 in a browser.

Navigate to the Play tab of the frontend. (For the moment, disregard other tabs as they are not completely up to date.) To test the system, try the example from the paper: