Using social sentiment to reduce cyber-bullying through text messages

In 2015 a study in the United States found that the average age of kids when they received their first cellphone was six years old. Also in 2015, almost 2,000 mobile apps were submitted to the Apple App store per day. Being a part of social networks and communicating through mobile devices is reaching a younger age each year. Over half of adolescents and teens have been bullied online and about the same number have engaged in cyber bullying. More than one in three young people have experienced cyber-threats online. Well over half of young people do not even tell their parents that cyberbullying occurs. Cyberbullying not only causes problems with social development, academics, but has even caused an increase in child suicides.


Current technology in the market, that has been developed to monitor communications, has been focused on after the text is sent, after the post is posted, etc. These software solutions help parents track what their child is saying or information they're receiving, but the text, post, content is already out there. These applications act like spyware, where this project puts the decision in the child's hand. It gives the child the opportunity to make a better choice by recommending more positive content for them to text or post.

2017 East Village, NYC

Finding the business value:

I've been working on this project for a few years on the side - iterating the concept, shifting technology, working through different use cases.

1. I originally designed the solution to be focused around tracking individuals language sentiment and assigning positive and negative values to words so individuals could see trends in their tones.
2. Next, it evolved with a social focus. The individuals could see their friends sentiments - who trended nice and who trended naughty.
3. Then, it took a turn to focus around family dynamics. Family members could see the broad sentiment of each other and be able to intervene if they noticed anyone trending to negatively.

Current iteration:

The solution would be in the form of a mobile keyboard extension, ie. Gboard. Accepting the keyboard extension terms would enable it to take in text message data with no tracking. This is not meant to be a spy software for parents. This application is meant to show young adults the sentiment of their language and keep them from sending hurtful text. The keyboard extension would have all the functionality of a regular keyboard, with Watson Tone Analyzer added as a top layer. As the child types, the sentiment changes. If the sentiment is too angry, based on a set threshold, the send button is disabled. Thresholds can be set based on age, language practices in different cultures, etc. Text sentiment is shown via a visual progress bar. Recommendations are given based on similarities of content. The recommendations are provided in a more positive tone.

Showing sentiment

Read left to right.

Providing recommendations

Read left to right.

10 second demo:


Submitted a patent through IBM in July 2017, it passed the initial committee, but didn't pass the search. The patent application didn't have enough differentiation from existing patents. I am currently working on a second draft to patent a potential aspect of this idea, not the entire idea.



ux design mobile app health