|
|
|
|
|
|
by feefie
4037 days ago
|
|
|
If Medical Watson works for and is paid by cancer hospitals, does that affect Chef Watson's recommendations (either in a positive way, to keep us healthy, or in a negative way due to the profit incentive)? In other words, do the various Watson implementations have any shared data or code base that could influence the other? I assume not, but how cool would that be? |
|
|
Most of the different Watson tools share various code elements - for example, we're not writing a new system natural language processing, entity recognition, and relationship extraction for each tool. Watson technologies, however, are designed so they work inside of a particular domain or with a specific corpus. So, the work with MD Anderson (http://www.mdanderson.org/newsroom/news-releases/2013/ibm-wa...) has a nearly 100% separate data set from Chef Watson with Bon Appétit. I say nearly 100% separate data set because there are some common data sets derived from publicly available sources (e.g. wikipedia) that may be used by both systems.
One of the other challenges is who the systems are designed for. Although anyone can use Chef Watson, it's not designed with everyone in mind. We've realized that the people with the best experiences are a little bit adventurous when it comes to cooking - they'll play around and improvise with flavors to create new recipes. This is a different demographic than would likely be using the health care system (doctors and care providers).