Using AI to Analyze Airport Reviews on Facebook

  3 m, 2 s

At a party this summer, a friend of mine and I were talking about the traveling we do for work, specifically the various hazards and hidden delights of America’s airports. Others at the party overheard us, piping in with their own ideas about which airports were “the best” and which were “a nightmare.” Most folks offered their opinions based on their own anecdotal data.

This discussion stayed with me. Not because I have an obsession with airports or because the party was so boring that airport chat was the highlight of the night. No, what kept the question with me so long was that I knew I could find out the answer. Using data and text analytics, I knew Semantria could mine online reviews about the biggest airports in the United States to find out which ones really are the best and most-liked.

While this “shower-thought” might just exit someone else’s mind as quickly as it entered, I decided to run this analysis as an experiment. As a first step, I decided to look at the ten busiest airports in America. I evaluated reviews for airports in Atlanta, New York, Dallas, San Francisco, Phoenix, Las Vegas, Los Angeles, Houston, Denver, and Chicago. Using a sparse data set, we discovered some very interesting things.

Comparing airports based on customer insight is not a new idea, but thus far the results have been based on individual subjective opinions rather than quantifiable data. There are hundreds of lists grading these airports published online and in print. There is no major consensus among these subjective articles as to which airport is considered the “best,” though Phoenix Sky Harbor International Airport often comes away with the top spot.

Regardless, my analysis delivered decisive results. For example, San Francisco’s airport rarely makes any of the airport quality listicles. If it does, it’s often not found in the top ten rankings. Despite this, travelers praise it as one of the finest airports in America for a bevy of reasons. The analysis I ran in Semantria for Excel turned this unexpected result in just minutes.

This creates an interesting question. Why is San Francisco’s airport, for example, so well reviewed by its customers while the travel industry virtually ignores its favorability? Similarly, why is Phoenix more highly-rated by travel industry reviewers than from the customers themselves? The forthcoming analysis succinctly answers those questions, and more.

When I initially ran these analyses, I didn’t apply Lexalytics’ airline industry pack. Interestingly, in every analysis I conducted “security” rated positively. This surprised me because, at least in my experience, getting through security at the airport is a miserable process at best.

After turning on the airline industry pack configuration, security went from bright green to dark red—the sentiment weight dropped from a net positive to a net negative. Enabling this industry based configuration is as easy as selecting the expiration date for your credit card… you literally pick it from a list. The airline industry pack allows me to see the conversation within the unique context of the airline industry.

Stay tuned, in the coming days I’m going to go alphabetically, airport by airport. When processed through Semantria, the data tells its own story, something deeper and more nuanced than star ratings and editorialized listicles. We’ll use this technology to take a deep dive into the collective voice of a large customer base. What they have to say is interesting, intense, and useful. First up, the busiest airport in the world—Atlanta.

Want to buy a print of the featured ATC towers in the header? 08left has you covered here.

Categories: Sentiment Analysis, Social Media, Special Interest, Text Analytics