Please log in to watch this conference skillscast.
Wefarm, the world’s largest farmer-to-farmer digital network, lets small-scale farmers connect with one another to solve problems, share ideas and spread innovation — all for free. With Wefarm, farmers can share and access crucial crop and livestock information about how to improve their yield or battle a disease without leaving their farm, spending any money, or having any access to the internet.
At Wefarm, Clojure and Natural Language Processing (NLP) has been used to design a service that works over SMS – in regional African languages not supported by any standard library – and that is robust to plenty of variation in spelling and punctuation. The flexibility of Clojure has enabled the build from a legacy system that only understands simple commands, to a conversational interface capable of understanding the intent and content of messages, and responding appropriately.
During this talk, you will discover models (Bayesian and elsewise), approaches (intent-matching with patterns written in a subset of Clojure), and tools (mini-REPLs) that have been used along the way.
YOU MAY ALSO LIKE:
Data Science in Languages I Don’t Understand
Rob is Head of Data at Wefarm, using Clojure, R, and Data Science/Machine Learning to help farmers get improved access to the vital information they need.