We did not have a concrete idea of what we were going to make going into the hackathon, but after learning about the HPE Haven OnDemand APIs, ideas started forming.
We were very interested in HPE's machine learning and text analysis API's and we collectively thought of an idea for a web application. Our idea was to create a service where emails can be analyzed for words that may be offensive towards women.
The Game Plan
With our idea in place, we drew out the design on a whiteboard to conceptualize what a prototype of our application would be. We wanted the user to be able to copy and paste their desired email into our platform. Then the user would be able to click on a button to display the modified email with offensive words highlighted on a textarea below. We found HPE’s Highlight Text API to be incredibly useful to identify and highlight the offensive words.
We initially supplied a list of over 100 words that were deemed to instill feelings of sexual harassment. We realized that we needed a database to store these words, and we came to the conclusion that MongoDB was the best technology for this purpose. We chose MongoDB because we didn’t need a SQL database and MongoDB was very quick to set up for a two day hackathon.
We realized that there could false positives so we wanted to provide users with the ability to add and delete words. Our database implementation allows us to easily add and remove words. This add and remove feature is aimed to provide HR managers the ability to add their own words that they wanted to flag.
Now that we have a way to analyze emails for offensive words, we wanted to add more insights about the email for users to see. We found HPE’s Sentiment Analysis API to be a great tool to identify whether the email was positive, negative, neutral, or mixed. In addition, the API also identifies the score detailing how negative or positive the email was. We decided that displaying the sentiment and it’s score along with highlighted offensive words would give an even more complete view of whether or not the email is appropriate.
With both highlight words and sentiment analysis features implemented, we then focused on improving the usability of the web application. We decided we needed to make the whole process of analyzing emails and sending them much more simpler. The first feature we implemented to streamline the email analyzing and sending process was to add a “Copy to clipboard” button, to easily copy the analyzed email to be pasted inside the user’s email client.
However, we wanted to take it one step further, and bypass the need for users to open a separate email client to send their email. So we implemented Oauth authentication for users to sign in with their Gmail accounts. Once logged in, the user can click the “Compose” button to open up a modal with the updated email formatted and ready to be sent. The user could also do last minute edits to change their email if they wish to (Perhaps there was no specific offensive words highlighted, but the sentiment score revealed that their email was unintentionally negative).
We managed to finish all the core features before the submission deadline was over. We then spent time preparing and getting ourselves ready for the pitch. We presented our pitch and our live demo of the application went smoothly.
Finally, it was time to announce the winning teams. We couldn't believe our eyes when we saw "Cleanify Email" on the projector, winning 1st place for "Best use of HPE Haven OnDemand to improve women's safety, health, culture, or economic empowerment". A moment of excitement and happiness came over us, as we went up to the stage for a team photo with HPE Haven OnDemand’s representatives.
We worked hard, had fun, and worked well as a team. We focused on developing quickly, and being able to pivot if needed. Finally, we would like to thank HPE Haven OnDemand and AngelHack for this amazing experience and opportunity.
If you would like to check out our code, CleanifyEmail’s GitHub repository link can be found here.
If you would like to try CleanifyEmail, it is deployed here.
From left to right: Martin, Thang, Spencer, Andrew, Tim
About Martin, Thang, Spencer, Andrew, and Tim
You must be a registered user to add a comment here. If you've already registered, please log in. If you haven't registered yet, please click login and create a new account.