Join us for Code @ Think 2019 | San Francisco | February 12 – 15 Register now Limited availability
Artificial IntelligenceDeep LearningMachine LearningSpeech and Empathy
By Rosie Lickorish November 5, 2018,November 5, 2018
While it might be obvious what physical attributes make a good prop or fly half, have you ever
thought about what personality fits best? Using IBM Watson™, England Rugby in collaboration with the IBM Research Emerging Technology team, were able to explore just that. With the goal of creating an engaging quiz for their fans, England Rugby used IBM Watson Personality Insights to analyse tweets from 51 England Rugby players, roughly 48,000 words, to find the patterns and trends among those playing in the same positions. The interactive quiz lets fans see which position matches their own personality, helping them connect with the England squad in a new and engaging way.
A well-accepted theory of psychology and marketing is that human language reflects personality,
thinking style, social connections, and emotional states. The frequency with which people use certain categories of words can also provide clues to these characteristics. IBM Watson Personality Insights infers personality characteristics from textual information such as tweets, social media interactions, and other digital communications. Using a machine learning algorithm that is trained using scores from surveys that were conducted with thousands of users, the service infers a personality profile with characteristics including the Big Five, Needs, and Values.
With the service available as an API on IBM Cloud, it only took a few minutes for IBM to analyse the personalities of the England Rugby squad based on their tweets. The results showed a range of characteristics within the team. Particular attention was given to the Big Five, one of the most
widely used and trusted of the personality models that can describe how a person generally
engages with the world. The Big Five dimensions are:
After the personality data from each player was gathered, it was used to create a collective view
for each of the 10 positions on the pitch: Prop, Hooker, Second Row, Flanker, Number 8, Scrum
Half, Fly Half, Centre, Wing, and Full Back. While no two positions showed identical personality
profiles, there were some similarities. From these 10 positions, 9 distinct personality profiles were
identified with various characterises. Interestingly, the profiles for Fullbacks and Wings were
very similar and so were grouped to form one profile.
The important value to consider when identifying each personality profile for the playing positions
was the percentile ranking. The numerical value between zero and one for each personality
characteristic as compared with a sample population, and if the ranking was average, high, very high, low, or very low. For example, a percentile of 0.91 for a characteristic indicated that the author’s writing exhibited the characteristic to an extent that was greater than 90% of the sample population and the overall characteristic was, therefore, considered to be very high. This approach
helped to simplify the results from each multi-dimensional personality profile and made
comparison between different positions less complex.
After the personality profiles of the 10 positions had been analysed, the Emerging Technology (ET) technical team worked with an IBM media partner, The Telegraph, as well as sport psychologist Dr Iain Greenless from the University of Chichester to draw out the interesting patterns and trends from the data. As such, this was very much a collaborative effort between journalists, psychologists, and software developers all working together to draw out the detail in the data. It was a collaborative effort between The Telegraph and the ET team who provided support examining the data details.
The final step undertaken by The Telegraph was to create the 20-question quiz to determine a fan’s
personality based on a well-known personality test. With their personalities assessed, fans can
find out where in Eddie Jones’s line-up their personality fits best!
Try it for yourself: englandrugby.com/rugbyquiz
The Watson Personality Insights service has arguably become a crucial tool across a range of industries, organisations, and businesses. Another example of its use in the sporting world is the Canadian basketball team, the Toronto Raptors, who use Watson Personality Insights in the assessment of team chemistry and to help weigh potential new recruits by analysing social media accounts, interviews, and more.
Beyond sport, IBM and other researchers have validated many hypotheses relating to real-world
applications of personality characteristics. For example, identifying which personality characteristics are more responsive to tweets, marketing campaigns, and predicting consumer preferences. Personality also influences interaction preferences between professionals and customers, including those between doctors and patients. For example, patients with a high degree of conscientiousness and openness where found to prefer being actively involved in deciding their course of treatment (Flynn and Smith, 2007).
Another key benefit of Watson Personality Insights is the ability to understand customers in far greater detail with the goal of creating personalised experiences. This can enhance a customer’s engagement and connection with a brand. The characteristics identified by Personality Insights can be mapped to user segments to create powerful, personalised content for different types of users, helping to form a deeper emotional connection with customers. Taking this one step further, it’s easy to image how other Watson services such as Sentiment Analysis and Tone Analyzer could be used to personalise a digital experience around a user’s tone, mode, feeling, emotions, attitude, or intent.
One key challenge faced by IBM and England Rugby when developing the quiz was the quantity of data available. While some of the players analysed were avid Twitter users, others were much less so. This
limitation restricted the pool of players that could be analysed from 58 to 51, as Personality Insights needs at least 100 words to provide a profile on an individual’s personality.
For the 27% of players who had tweeted between 100 and 600 words (on average about 40 tweets), Personality Insights was able to analyse the data with an acceptable level of accuracy. For the remaining 72% of players, Personality Insights returned results within three percent of the best “Mean Absolute Error.” This meant, that for the majority of players, the results returned from the service were very close to the scores the authors would receive by taking a personality test.
Using a relatively small data set, England Rugby was able to use Watson Personality Insights to
create a compelling and engaging way for their fans to get a step closer to the action. To find out what position your personality is most like, head over to englandrugby.com/rugbyquiz to have a go at the quiz powered by IBM Watson Personality Insights.
Interested in learning more about IBM Watson Personality Insights? Take a look at this sample
code and explore the following links:
Apache SparkArtificial Intelligence+
Back to top