Simple chatbot that interacts with google calendar. The code consists of 3 main parts: a chatbot section, an nlp section and a google calendar query API section.

The chatbot setion using google speech recogniton to listen the to users input and also to speak to the user (by converting text to speech). The chatbot currently supports English and French. The chatbot, which is called Alina, can be awakened using any of the following phrases "Hello Alina" or "Bonjour Alina" (for french). Based on the wake-up phrase, we choose the chatbot sets the default language it expects to be either English or French. The chatbot can ended be using any phrase that involves "bye" or "Au Revoir".

The chatbot listens to any user and converts any input phrase to a text (string). The spoken text is then passed to the machine learning model, which tries to identify what is the user's query. For this purpose, we used a multilingual model pre-trained on 15 languages to encode the input sentence from the user. This sentence is then compared against our database's English and French encodings of various sentences. Each of these sentences represents a query that the assistant can perform. We use a cosine distance as a similarity metric between the new sentence encoding and all the sentences in our database. The query for the sentence with the maximum distance is selected. We also set a threshold only to validate the query if the maximum distance is at aleast 0.5 (as the cosine similarity ranges from 0 to 1).

Finally, the selected query is implemented using functionalities from google calendar API. The API requires the user to provide a credentials file to allow the app to access their google calendar. Furthermore, a little bit of string manipulations and regular expressions (handle by the dateutil package) is required to capture dates and time information from the input sentence. The queried information from the google calendar is then passed to the text-to-speech functionality of the chatbot to respond to the user.

Create and activate a new virtual environment (recommended) by running the following:

python3 -m venv myvenv
source myvenv/bin/activate

Clone this repository and open a terminal inside the root folder:

git clone https://github.com/arnold402/NLP_project.git

cd NLP_project

Install the requirements:

pip install -r requirements.txt

Download pretrained model for encoding:

wget -O model/lid.176.bin https://dl.fbaipublicfiles.com/fasttext/supervised-models/lid.176.bin

Run the app:

python Bot.py