Girl Guides Canada - Sentiment Analysis App
I was apart of the Myplanet now Orium 2017 fellowship class which gave 18 students a real, hands-on experience with agile, as well as the ability to implement AI and the latest development platforms to build a product for a non-for-profit/charity.
As an added bonus, Myplanet partnered up with IBM meaning that we also had the ability to connect with some of the latest in cognitive tech while maintaining a community-oriented charitable component.
We spent the first week doing focused skills training — learning new coding languages and perfecting our understanding of certain approaches to development — and the second week working in teams to solve challenges for one of our three charity partners using cognitive-based technology solutions. Being it was my first development internship I would definetly recommend it for anyone in the Toronto area looking to get hands on experience as the mentors and people I met were extremely smart and passionate about learning new technologies.
The theme of Fellowship this time was Cognitive for a Cause. It allowed us to explore our joint interest with IBM in the real-world possibilities of cognitive tech (using IBM Watson) by applying it to key community initiatives. We specifically worked with charities that might not have the resources to explore the more cutting-edge or experimental ideas that cognitive tech is making possible these days. Yashar Rassoulli
Our team’s problem statement
GirlGuidesCanada is embarking on a massive change which will impact all of its members in some way, shape or form (~100k). Girls, parents, Guiders or other Sr. volunteers have no centralized way to share what they feel/think about the change and we have no way to hear, analyze and consolidate ‘the pulse’ of the field. We need a better way to get at this information to help us make decisions in real time!
Our solution
Our goal was to create an application that allows the client to gather data from social media.
Working in a team of 4 we built a real-time twitter feed prototype app that allows users to filter by key words, tags or twitter handles for the Girl Guides of Canada. We focused on utilizing one social network that being the Twitter API however the application could be scalable to other networks (Facebook, Linkedin, Instagram etc). In particular our twitter API data from these five hashtags:
#GirlGuides
#GirlGuidesCanada
#GirlGuidesCookies
#GirlScouts
#MyPlanet
In order to enable the ability for the client to be able to hear people’s thoughts and opinions on this change, we implemented filtering of tweets based on key words, hashtags or twitter handle. Liaising with IBM to incorporate their latest technologies into the project we analyzed the sentiment (joy, fear, sadness, disguist, anger) of each tweet body using the Watson Tone Analyzer API.
Source Code
- Stack used: React.js, MongoDB, Twitter API, Watson Tone Analyzer API