BUILDING A WEBSITE FOR COVID-19 INFORMATION SEARCH

Abstract

In this thesis, I construct a website to support looking up information, news, and statistics about the Covid-19 pandemic. Website is built on Django web framework and MySQL database. Besides, I integrated into the website a chatbot system using a retrieval-based method with a closed domain and short-text conversations about Covid-19 pandemic. The chatbot is built on RASA platform and uses several models such as Vietnamese Tokenizer for words separating, BytePair Featurizer for words processing, DIET classifier for extracting intents or entities. Moreover, I manually construct a chatbot corpus with 3 entities and 280 examples divided into 13 intents. The final result shows that the chatbot responses well with a 0,996 F1 score.