Abstract
For the last three years, our chatbots have come in 1st twice and 2nd once in the Loebner Prize Contest, with a different persona each year (Suzette, Rosette, Angela). Suzette even fooled a human judge. A world-class chatbot should tell the story of its life, have a consistent personality, and respond emotionally. It takes a lot of script. And it takes a powerful engine designed to support natural language processing in a variety of ways and make it relatively easy to author all that script. This paper briefly discusses ChatScript, the open-source Natural Language scripting language and engine running our bots. Then it looks at how we construct chatbots and what we have learned.
Highlights
RESUMEN: Durante los últimos tres años, nuestros chatbots han conseguido dos veces el primer puesto y una el segundo en el concurso Premio Loebner, con un personaje distinto cada año (Suzette, Rosette, Angela)
That code can sions I’ve noticed myself so immersed in the expe- loop, execute conditionally, invoke topics
Rejoinders are associated with a specific rule and are matched against user input only when that rule generated the most recent chatbot output
Summary
The task of a chatbot is to create an illusion – the its output script and rejoinders as its own. The pattern in parens specifies when a rule is derstands and cares about what you are saying. At a time, wrote about the final version: After the pattern is the output. The talk started brilliantly, so light and witty, I lo- is just literal text to say. On a few occa- plex code with or without output text.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have