Developing a Chatbot with RASA — Part 1

Part 1: Installing RASA

RASA chatbot

This article is aimed to help understand the basics for installing and setting up RASA Open Source on an Ubuntu machine. At the time of writing this article, here are the various versions of the required software:

1. Python = 3.8.3
2. RASA = 1.10.5

The RASA installation steps are being followed along with some of the observations. The RASA Open Source currently uses Python 3.6 or 3.7, hence we will first enable a virtual environment and use it as our working directory.

Pre-Requisites:
a. pip > 19.0
b. Anaconda (preferred) or Virtualenv
c. >5 GB space
d. sudo access (rights)

** At places below, due to some constraints, “-- ” is shown as “— ”. So please re check the same if you face an issue.

Step 1: Create a Virtual Environment
a. using venv:
python3 -m venv /path/to/new/virtual/environment

b. using conda environment:
conda create — name py37 python=3.7
conda activate py37

Step 2: Install RASA
pip install rasa
*if you have pip3 as default, use: pip3 install rasa

you may get some Errors where some of the libraries may not be installed as they might be of higher version.

Errors while installing RASA due to incompatible library versions

To tackle this, install the library for that specific version suggested, using pip. This will uninstall the higher version and install the version mentioned by us.

pip install scipy==1.4.1
Finally, before moving to next step, check the RASA version
rasa - - version

Step 3: Initiate a RASA project with defined structure.
rasa init- -no-prompt
The above command will create a default RASA project and sets the directory with basic folder structure.
Also, with the same command, Training of the model will also happen.

Step 4: Run the shell to initiate the Chatbot
rasa shell
The above command will initiate the Chatbot and we can then start interacting with the Chatbot.

Before we wrap this article, here are some key points to observe:

1. Having sudo access helps in smooth installations
2. Since the versions are frequently updated, default libraries’ version can keep on changing. Having the Virtual Environment can help install the required version.
3. Follow official documentation and if you still face an issue, raise the issue on the RASA Forum.

I hope this article helps you overcome the challenge you might face to get started of RASA.
Stay Tuned for next article, which will be on the Components of RASA.

Connect with me on LinkedIn or Email for further discussion on Chatbots and or RASA.