Flyte

A real-time social media sentiment analysis tool

Comment

Inspiration

In recent years, the rapid growth of social media and the ability for businesses to use this tool to promote their products and events has lead to the outpouring of hundreds of thousands of dollars into online campaigns aimed at sparking positive public interest in such efforts.

Until now, if businesses wished to analyse the effectiveness of their campaigns they had two options; to see whether their event or product was trending, or to write their own sentiment analysis software. However, these solutions tend to be either ineffective in giving detailed analysis of the campaign, or costly to the business.

Flyte offers an alternative to these options, and provides a real-time analytics tool to effectively monitor public consensus on social media. Through examining the sentiment of tweets on a user-defined topic, businesses can access graphs, charts, and data visualisations that determine what is (and what isn’t) working in their campaign.

How it works

Flyte is a real-time social media sentiment analysis tool. It uses the Twitter stream API to help understand the public consensus towards topics and issues online.

Our application will gather relevant tweets towards for a given query, analyse its sentiment (positive, negative or neutral) over a repeating interval and display how a given topic is currently being viewed online.

Challenges

  • Handling multiple users: The Twitter Stream API can only be used on a single application per key. Given that we wanted our application to serve multiple users we needed to come up with a way to serve multiple users with a single stream. We decided to have all queries pass through the stream, then we would filter out tweets depending on the users query.
  • Avoid API quotas: Given that we are doing a real-time analysis, api quotas can usually be an issue. We were able to avoid any issues by making use of HP Batch API calls which allowed us post data 1/5s.
  • Extracting useful information from a sea of raw data.

Accomplishments

  • Learning new technologies
  • Seamlessly co-ordinating our data retrieval with our visualisations, to accomplish our project goal (creating a real-time web-app)
  • We genuinely believe that this project is beneficial for many people such as: small businesses looking to manage their online, social media profile.

What we learned

  • Despite our scope of ideas being somewhat broad, we had to focus our view on what was relevant to our goal.
  • Adapting to unexpected allowed us to grow as developers in a high pressure situation

What's next for Flyte

  • Ability to view multiple graphs at the same time, to analyse how one trend may affect each other: English Cricket Team vs Australian Cricket Team
  • Allow users to enter their own time span and process graph of the the sentiment from this timeline.
  • Migration from HP’s sentiment api, towards custom built software. Why? Accurately train on data on different social networks and localities.
  • Analyse trend in sentiment to predict events before they happen eg. Riots etc.
Share this project:

Updates