Tesla's Twitter was hacked a few weeks ago reminding us that social media accounts are ever vulnerable to nefarious compromise. Such events are damaging to the brand because of the radical shift in online presence. But a damaging tonal shift can just as easily come as a result of staffing changes in a brand's PR team or a new advertising campaign.

How it works

ProfilePug periodically pulls user-generated text from social media profiles to plug into Watson's Personality Insights; starting with when the user first signs up for the service. ProfilePug compares the user's personality profiles at multiple time points to assess the likelihood that the user's account is being used by someone else.

Challenges I ran into

Tweets tend to contain plenty of text that is likely not representative of the user's personality. Parsing the content of tweets in an intelligent way proved particularly challenging. Collecting enough data from tweets for Watson's Personality Insight's to work properly was another challenge since the size of the confidence intervals is inversely proportional to the number of words supplied for analysis.

Accomplishments that I'm proud of

We are proud our ability to quickly leverage the Watson platform to make using the internet a more worry-free experience for individuals and brands alike.

What I learned

Learned about the mechanics of Watson's Personality Insights. Learned about a few different methods of scraping a Twitter feed.

What's next for ProfilePug

We'd like to adopt the basic premise to other social media platforms and blogs.

Share this project: