Inspiration
We were inspired to create this product after our team realized that a lot teens in America face cyber threats or cyberbullying to some extent. We realized that not a lot of support is around to stop these threats or even detect them in the first place, so we decided we would step in and offer a solution.
What it does
What the cyberbully detector does is it takes text that the user inputs and runs it through an algorithm that we created which consists of metrics and thresholds that the program runs against. We paired up google clouds sentiment analysis with some profanity checkers and some of our own code to create a backend that will detect cyber bullying to the best extent it can.
How I built it
The way we built the product was by first learning about the documentation and processes that went into Sentiment Analysis on googles cloud platform. From there we went ahead and tried to create our own machine learning model that would detect sentiment and emotion in text as well. After we were able to get that up and running, we decided to pair this up with a profanity checker to make sure that if a user does showcase profane language in the text, we would be able to detect it and make a stronger case that the texts were cyberbullying incidents. We then incorporated Django as our framework as it was a powerful tool that we have used before and had most familiarity with. Finally we decided to create our frontend using HTML and CSS where are able to show the quantities that occur from our threshold and metrics algorithms. We were also able to get the colors to change depending on the mood of the text.
Challenges I ran into
The challenges that we ran into was mainly getting started and setting up the frameworks and toolsets. It was our first time working with a lot of these tools so the learning curve was a little bit high but we were able to overcome it and implement the designs to the best of our ability.
Accomplishments that I'm proud of
My team and I are extremely proud of how we were able to overcome challenges such as creating a machine learning model and not giving up when it seemed bleak. There were times when we thought it was not worth it but thankfully we stuck through and created a great product.
What I learned
We learnt mainly about google cloud platforms machine learning models.
What's next for Cyberbullying Detector
Hopefully we plan to incorporate a way so that the user can see what part of the text is more emotionally charged in comparison to others.
Log in or sign up for Devpost to join the conversation.