A chatbot is an intelligent piece of software that is capable of communicating and performing actions similar to a human. Chatbots are used a lot in customer interaction, marketing on social network sites and instantly messaging the client. Our chatbot is a COVID CHATBOT. The primary focus of the bot is to guide people in the ongoing pandemic, answer questions regarding vaccinations, symptoms, live covid statistics, probability of being infected to name a few.
To run the chatbot locally: python3 chatgui.py
chatgui.py
contains the GUI of the chatbot using which the user can interact with the bot. Tkinter
has been used to make this.
- Run
python3 train_chatbot.py
in the terminal to train the chatbot.words.pkl
,classes.pkl
andchatbot_model.h5
will be generated.words.pkl
stores the the words Python object that contains a list of our vocabulary,classes.pkl
contains the categories andchatbot_model.h5
contains the trained model that contains information about the model and has weights of the neurons.
- We have trained our chatbot using the
keras
module andnltk
wordnets to the intents.json present in the Data folder. - A special recurrent neural network (LSTM) is used to classify which category the user’s message belongs to and then we will give a random response from the list of responses.
An existing dataset (https://github.com/nshomron/covidpred) was used to collect data regarding 7 features and the state of the patient(i.e. if she/he has COVID or not). Using the data, a Naive Bayes Classifier from the sklearn
module was trained in order to predict new cases of COVID. The user is asked to input the details in a specific format after which the classifier predicts probability of the user having COVID.
The statistics for the mentioned state are being scraped live from https://www.oneindia.com/coronavirus-affected-cities-districts-in-india.html using BeautifulSoup
.
News regarding covid for the mentioned state are being scraped live from Bing news using BeautifulSoup
.
The bot tries to answer questions regarding vaccinations and directs further questions to the cowin website of the Government of India.