Agenda:
19:00 - Welcome from the organizers
19:10 - "In Search of Best Practices for NLP Projects" by Ivan Bilan (TrustYou)
19:30 - "Genie, An Open-Source Toolkit for High Quality, Affordable Virtual Assistants" by Giovanni Campagna (Stanford OVAL)
19:50 - "Conversational Privacy: How can chatbots and speech assistants communicate privacy?" by Birgit BrĂŒggemeier (Fraunhofer IIS)
20:10 - Hangout
Topic summaries:
"In Search of Best Practices for NLP Projects" by Ivan Bilan
In this talk, we are going to explore the best practices for building NLP and ML projects from the ground-up and up to production. What does it mean to build a prototype? Do I really need CI/CD for my NLP projects? What does my team need to know to work on NLP Projects? Anything we can learn and use from ML and DataOps best practices? How to stay on top of recent NLP developments? These and many more questions we will explore and discuss together.
"Genie, An Open-Source Toolkit for High Quality, Affordable Virtual Assistants" by Giovanni Campagna
Virtual assistants like Google Assistant and Alexa are a fast
growing market. In this talk, I will introduce Genie, our open-source
tools that allow companies and communities to build virtual assistants,
reducing the reliance on large companies and improving privacy. Genie
uses neural semantic parsing to translate natural language to the
ThingTalk programming language we designed for this purpose. We
introduce synthesis to reduce the data acquisition cost. Results on
several benchmarks show that Genie is effective for complex questions
and task-oriented dialogues, at a fraction of the cost of the
alternatives.
