By Elena Marchesi & Ahmed Taha, 2019
The focus of this project was on the current “SciComm Chatbot”. In fact, working with an already existing project, initial researches to acknowledge its state were needed to develop an improvement strategy.
Once we familiarized with the current “SciComm Chatbot”, its main weakness was the absence of user tests, fundamental to highlights flaws in the chatbot.
A preliminary hypothesis on the main risks of the chatbot had been made, including:
• Chatbot not working correctly, not answering to an input;
• Problems with the database (too big, too small or not including most important Q&A);
• Conversation that feels fake or robotic;
• Keywords not recognized (language risks).
All these risks can be mitigated through continuous users testing. Therefore, our project will be conducted using and an experimental approach by testing at multiples levels and steps.
In addition, whenever the chatbot reveals to be successful on Facebook, an initial step towards its application on the SciComm website can be taken.