2019-09-15 –, Ferrier Hall
Now we can talk to our gadgets as if it’s a real human. It makes you wonder when you chat with the “customer service” online, is that a real person or a robot on the other side? I built a chatbot, and here’s 3 things that I discovered.
Chatbot is cool, have you even think about building one yourself? There are many platforms and tools available, do you know which one to use and where to start?
The first thing I am going to tell you, is what are the main components of a chatbot. It’s always a mystery why chatbot can understand us and take action accordingly. To understand such a complicated machine, we will tear down a chatbot and show you the different parts of a chatbot: NLU, dialogue logic, user interface; and explains what their functions are, thus which part of the process they are responsible for, and how they interact with each other.
The second thing, what a chatbot can and cannot do. We always have an impression that all chatbot can pass the Turing test easily due to the advancement of deep learning and AI. Is that true? Or is it just a couple of big key players in AI showing off their ability? By showing how a chatbot is trained, we can see what limits the “intelligent” of the bot. We can also see how we could make adjustment in training to suit different purpose.
The third thing, is which one to choose. There are many different chatbot development platforms available, both commercial and open source, and they are different in design and approaches in training the bot. I have tried using Amazon Lex and Rasa. From my experience, I can tell you what are the difference between using them so you have a better understanding of what suits your use case better.
This talk is suitable for those who are curious but does not have much experience in chatbots. We will assume audiences already have basic knowledge about NLP and neural network so it would not be covered in details.
Cheuk constantly contributes to the community by giving AI and deep learning workshops, organize sprints for open source projects, volunteering at Datakind for charities. At the same time contribute to open source projects including Pandas, Keras, Scikit-learn and Dateutil. Cheuk has also been a guest speaker at University of Oxford and Queen Mary University of London, and various conferences including PyData in Amsterdam and Berlin, PyCon in Israel, UK and Germany, EuroPython and PyLondinium. Believing in gender equality, Cheuk is currently a co-organizer of AI club for Gender Minorities to support Tech Diversity and Inclusion.