Lifty++ is a Google extension that aims to create a more positive experience on social media for all types of users. According to #HackHarrassment, 40 percent of users have experienced online harassment in some form. To discourage this type of content, our project aims to warn users when they may be using harsh language. If their text is over 30 percent negative, we will ask them to use more positive words.
How we built it
We used Python to develop the algorithm that would calculate the sentiment score of each word, and eventually everything written in the text. Negative words received a score of -1 while positive words received a score of 1. Neutral words were disregarded and assigned to a score of 0. We summed up the values associated with each word and calculated the percentages for each category by dividing against this total. We used the NLTK library to tokenize each word.
Challenges we ran into
We originally planned to involve speech recognition using Google Cloud API and the Youtube API but ran into several issues with compiling them into a chrome extension. We also ran into challenges with executing our Python ASCII code but ultimately overcame these issues.
Accomplishments that we're proud of
We built this extension from the ground up. We are proud of writing our own algorithm that categorized the words according to the sentiment.
What we learned
Dealing with ASCII characters was more difficult than originally planned.
What's next for Lifty++
Eventually, we plan to incorporate speech recognition into Lifty++. This could range from audio to videos, such as those posted on YouTube or Vimeo. We also plan to suggest word replacements for negative words rather than simply delivering a warning to the user.