USAF-Thunderbirds

CLIENT: United States Air Force Thunderbirds

The USAF Air Demonstration Squadron is the air demonstration squadron of the United States Air Force. The Thunderbirds are assigned to the 57th Wing, and are based at Nellis Air Force Base, Nevada. The client wanted a solution that can help them get insights on public opinion about their event’s on social media.

Objective

The objective was to analyze twitter and Facebook posts and generate insights related to public opinion on the squadron and events hosted by the squadron.

Challenges

1. Biggest challenge in creating this system was to get the data from the social media and categorize it before getting results. The data was unstructured text which had to be processed to get meaningful insights.

Process

The social media posts were retrieved using the APIs made available by Facebook and Twitter. The data was retrieved from posts containing specific hashtags. Hashtags allowed retrieval of relevant posts only.

Posts data was then processed using NLP to identify the content of the messages. By performing sentiment analysis, posts were divided into 3 categories:

  • 1. Positive
  • 2. Negative
  • 3. Neutral

Solution

The final application consisted of a mobile application that allowed users to:

  • 1. Take Surveys on social media based on hashtags
  • 2. View statistics of user posts with these hashtags
  • 3. Get insights on public tone about the topic

The mobile application used Facebook and Twitter APIs for collecting data. The application used IBM Watson for AI on the backend. IBM Tone Analyzer was used for Sentiment Analysis which provided details on following aspects of the post:

  • 1. Analytical
  • 2. Tentative
  • 3. Anger
  • 4. Confident
  • 5. Fear
  • 6. Joy
  • 7. Sadness

Using the above data, the application could also identify any threats that were communicated in the posts.

Technologies

Screenshots

Results

The application was successfully used for multiple events hosted by USAF Thunderbirds and insights were used to improve the events. The application was later modified to also get insights on the individual Thunderbird crew members