On July 8th, Rónan McAteer from the Watson Solutions Development Software Group in IBM Ireland will give a talk as part of the Big Data Information Visualisation Summer School here in St Andrews. This talk is entitled “Cognitive Computing: Watson’s path from Jeopardy to real-world Big (and dirty) Data.”
While this talk is part of the summer school, we are trying to host it in a venue in central St Andrews during the evening of July 8th so that people from across St Andrews and SICSA can attend if they wish.
The Abstract for Rónan’s talk is below:
Building on the success of the Jeopardy Challenge in 2011, IBM is now preparing Watson for use in commercial applications. At first glance, the original challenge appears to present an open-domain question answering problem. However, moving from the regular, grammatical, well-formed nature of gameshow questions, to the malformed and error-strewn data that exists in the real world, is very much a new and complex challenge for IBM. Unstructured and noisy data in the form of natural language text, coming from sources like instant messaging and recorded conversations, automatically digitised text (OCR), human shorthand notes (replete with their individual sets of prose and typos), must all be processed in a matter of seconds, to find the proverbial ‘needle in the haystack’.
In this talk we’ll take a look at how Watson is rising to meet these new challenges. We’ll go under the hood to take a look at how the system has changed since the days of Jeopardy, using state-of-the-art IBM hardware and software, to significantly reduce the cost while increasing the capability.