Joe Francis & Pietro Mazzoleni - IBM World of Watson - #ibmwow - #theCUBE
summary
An app recommends sessions from 1,200 choices by combining Watson Conversation, a secondary conversation instance, Cloudant, Bluemix and Node.js, built in about four to five months. The team is ~20 people (ten full time, ten on call) with UI, UX, backend and infrastructure roles. Machine learning retrains every few hours; effectiveness is measured by how many of the top 10 predicted sessions users actually pick and by live monitoring and performance alerts. Recommendation explains why and soft launch testing validate quality.
AI-written summary · about
