Watson predicts the winners: Eurovision 2017

  • Posted by admin on May 10, 2017

Yodeling rappers, dancing gorillas, a horse on a step ladder, excessive key changes – that’s right ladies and gentlemen, Eurovision 2017 is here, and this Saturday we’ll see who shines and sings to steal the show.

I’ll admit I am writing this post as an excuse to spend work time watching the rehearsals and reading up on the odds, favourites, and must sees. The fact that I am hosting a Eurovision party on Saturday is neither here nor there. Prep is prep, and if you can get paid for it, all the better.

But that’s not the reason for this post, at least that’s what I am telling my boss. I have always maintained if you want a crash course in Politics, Geography, Global Finance, and Music all in one you need to tune into the Eurovision Song Contest. It’s a Saturday night eye-opening and enlightening educational experience. And as part of this education we decided to throw in some Science, Technology and Psychology and see what Watson makes of this years’ entrants using the Watson Tone Analyzer tool. Will Watson predict this year’s Eurovision winner? Read on…

The Eurovision Song Contest

Don’t be fooled, it may be called the Eurovision Song Contest, but it is neither about singing, songs, or in fact Europe.

Hosted this year in Kyiv (Ukraine were last year’s winners) this year 43 countries will be taking part. There are two semi-finals (one was held on Tuesday night, the other one will be on Thursday evening), with the final being held on Saturday 13th May 2017.

The grand final is made up of 26 countries, ten from the first semi-final, ten from the second, plus the ‘big five’ (the UK, France, Germany, Italy, and Spain automatically qualify because they contribute financially towards the event), and Ukraine as the host country.

You won’t win based on the tune alone. It’s a balancing act between the show, the dry ice, the fashion. Between the countries, and locations, and alliances. Between the recent and not so recent politics.

What is Watson Tone Analyser?

IBM’s Watson Tone Analyzer uses linguistic analysis to detect ‘tone’ in written text. Businesses can use the Tone Analyzer service to help them understand conversations and communications, allowing them to respond to their customers better.

You could, for example, use it to monitor social media and other web conversations, including short form copy (like tweets) or longer documents like articles and blog posts. Or you could use it to monitor customer serve and support conversations at scale.

But we’ve decided to show some fun insights about Eurovision songs and lyrics. You can see how it Watson Tone Analyzer works, and, of course, be in a better place to pick your winner!

How does Watson Tone Analyser work?

  1. Go to Watson Tone Analyzer
  2. Click on ‘your own text’
  3. Open the copy that you want to analyse (in our case one of the songs) and copy it paste it into the Tone Analyzer
  4. Click ‘analyze’
  5. An overall summary for the document is provided
  6. Scroll down to view your text in more detail. Identify specific sentence where tones are the strongest

The patterns of past Eurovision winners

It is easy to see a pattern in the tone of past winners, with specific emphasis on ‘emotional tone’, and social tones of ‘agreeableness and ‘conscientiousness’.

Emotional tone measures different types of emotions and feelings that are expressed.

For each emotion, a score of less 0.5 indicates that the emotion is unlikely to be perceived in the content. Likewise, a score greater than 0.75 indicates high likelihood that the emotion will be perceived.

Our past winners all show either sadness, fear or joy in their emotional tone.

For social tone the tool measures social tendencies, using five categories (openness, agreeableness, conscientiousness, extroversion, and neuroticism) adopted from the ‘Big Five’ personality model (not to be confused with the Eurovision Big Five).

Past Eurovision Song Contest winners and their Watson Tone Analyzer scores

Past Eurovision Song Contest winners and their Watson Tone Analyzer scores

The past Eurovision winners we analyzed show high levels of agreeableness and conscientiousness. With the only anomaly being Sweden with low scores on both in 2012.

Watson’s favourites to win

Green indicates a clear match with the patterns of past winners. Orange indicates a close, but not conclusive match with one category not aligning with past winners.

The 2017 Eurovision Song Contest entrants and IBM Watson Tone Analyzer scores

The 2017 Eurovision Song Contest entrants and IBM Watson Tone Analyzer scores

Italy is currently the bookmakers favourite, but according to Watson it is way off the mark with low agreeableness and low conscientiousness.

The scores are ok for the United Kingdom. But based on past results it doesn’t bode well, having struggled with last places and nul points since their last win 20 years ago with Katrina and the Waves. I think it’s safe to say that the recent Brexit vote is unlikely to help. The biggest question is really whether the UK will get any votes, rather than will they be in contention as a winner (which Watson can’t predict).

My co-author Demetre is in favour of Portugal, but I frankly think this one is boring and forgettable, sorry Portugal! On the opposite extreme we’ve got the memorable (and quite catchy) rap yodelers from Romania. They still need to qualify for the Grand Final, but the Watson scores doesn’t rule them out.

Watson’s hot favourite is France – not a bad bet given the current political element either.

We’ll check back in on the second semi-final and give some updates then, but looking at the Watson numbers so far you should expect to see Belarus in the Grand Final, along with Croatia, and Ireland, all of which match the patterns of past winners.

Next steps

If you can’t wait to find out the winners, why not run your own report using Watson Personality Insights? From books, scripts, papers and manifestos, you can draw patterns from all kinds of written data.

The post Watson predicts the winners: Eurovision 2017 appeared first on Internet of Things blog.

Internet of Things blog

Leave a Reply

Your email address will not be published. Required fields are marked *