It was some time ago that a friend of mine suggested I might like a Watson experience. This week, I had some time on my hands and decided to take a peek using a bit of data I’ve gathered.
Curvaceous Learning
In wanting to get to know the technology prior to investment, I tried to logon to IBM’s site and found that there were no learning options available without a pre-credit submission. Nervous about being charged some fees, I scampered off to LinkedIn learning where I was talked into to taking the free trial plunge.
After the learning course, I side-stepped over into IBM’s learning resources. I was this video that finalized the decision to activate the trial and create an instance.
Along with this documentation.
The Journey
Now we discuss the journey that was taken into working with Watson. First there’s the bone shaking state of signing up for the free trial.
There are distractions such as this one. Do I want to take the course or do I want to follow the other documentation?
I returned to the original step by step documentation as I felt compelled to complete something.
The sample wasn’t good enough, so I navigated to the other tutorial.
The Distraction
The tutorial taught me that Watson can handle a compressed file, so there’s no need to unzip unless the compressed file you’re working with involved a Mac.
And this is what it looks like if you observe what’s going on as you’re waiting for the tool to complete the collection.
The UI/App portion of the tutorial required that I engage in a local repository This wasn’t part of today’s learning goal. I’m interested in data.
Kalika’s Preference Data
Imported Kalika’s preference data into Watson to see what I would glean:
Based on the question, the bot should return one of my most liked teas. That would be, the tea with the highest rating and/or the highest sentiment. Leaving ratings to the side, for now, as the goal is not to investigate Watson’s sentiment accuracy. I zoomed in on how to get my favorite tea to be the response to the question.
With bit of finagling, I learned that Watson doesn’t seem to get along with my multiple-sheet Excel document. I reduced the data that to Twining’s Tea.
I worked with various settings and came across this error.
Somewhere along the line I learned this solution and I can see some results.
Findings
Here are my findings after a brief stint with Watson. It was fun. The way that I have structured my data, this option doesn’t seem beneficial. I have to put a little more thought into this and determine what it is I’d like to achieve.
I can get the tool to tell me what kind of tea to buy, but I cannot seem to get it to tell me whether or not I like something:
There’s more work to do. Watson is a pretty fun little utility that I think performs better with Sherlock, not me.
I clearly have more work to do, but I don’t think it’s with Watson.
Leave a Reply