Healthy Online Social Track [HOST]

An attempt to avoid cyber bullying of a user through Machine Learning

Comment

Inspiration

According to several studies, cyber bullying has increased in these days of social media. Cyber bullying can happen due to a variety of reasons such as peer pressure, want of revenge, basic lack of empathy etc.,. It is extremely important to protect and shield the victims of cyber bullying as they run the risk of being psychologically damaged which can potentially affect their future.

What it does

Healthy Online Social Track (H.O.S.T) seeks to identify bullying messages on social media such as facebook, twitter and provide insights about them so as to prevent the reader from getting affected by those messages.

Our demo is a mobile application which take an input message from the user. This message is then analyzed by Google Natural Learning Processing and returns how positive/negative the user's message is. Based on the result, the user can reconsider whether or not they want to post this message online.

How we built it

Built using: Weka, Google Cloud natural Language API, Google Cloud AutoML API, Amazon Machine Learning - Predict API, React-native

Challenges we ran into

Data gathering and data preprocessing - Since this is about cyber bullying, we found it hard to find data regarding cyber bullying. Once we found the data, there was lot of redundant and noisy data, so it took a long time to pre-process and clean the data.

React-native - Due to the complex software foundation of react-native, we had trouble getting the application running on all of our individual hardware.

Accomplishments that we're proud of

Contributing to prevent/avoid cyber bullying in some way.

What we learned

For all of us except Shreekrishna, this was the first Hackathon and we gained a lot from this experience like building applications within a 24 hr timeframe and got the chance to interact with a lot of like-minded and talented people.

What's next for Healthy Online Social Track [HOST]

Next step is to implement Speech to Text API where in addition to text, the app will be able to find bullying messages even through speech.

Built With

Share this project:

Updates