watchfalcon, protect your kids from online threats.

A safety assistant on the phone for passively monitoring all communication in the background to protect kids from cyber frauds, predators and online abuse.

Comment

Inspiration

Problem: Kids these days are facilitated with mobile phone and the internet, and this has lead to rising in cyberbullying, adult content, sexual predators, profanity, and threats of violence against children through various communication or social media platforms.

Parents can not monitor kids phone regularly without breaking into a child's privacy. Often parents would need to spend hours every week to monitor a child phone and educate them.

What it does

WatchFalcon once installed on a child's phone, will listen to all of their communication channels in the background without affecting their privacy or their UX and analyzes the context and sentiment of the conversation using expert.ai's NLP APIs in realtime and send alerts to their parents only if expert.ai finds cyberbullying or sexual exploitation or suicidal or unethical context.

Innovation/USP: WatchFalcon works irrespective of any social media platforms, as it uses the View-Hierarchy Tracing algorithm along with expert.ai NL APIs on Android, which transverse through the content/text being shown on screen and converts it into dialogue-based sentences. Thus making our app automatically scalable to any new app that is installed on the kid's phone.

And as our app alerts only the danger content to the parent's, the children could have their own privacy.

Our app works like this:

  1. Install: Parents install the Falcon Tracker app on their kid’s phone and activate it.

  2. Monitor: WatchFalcon will start to extract and monitor text messages, emails, and social activity on all social media from Whatsapp to Snapchat for signs of harmful interactions and content.

  3. Analysis: It converts the extracted content into dialogue-based sentences from the conversation and uses expert.ai Document Analyzer API to detects/mark a conversation for digital dangers. Sentiments, main syncons, main phrases, and classification parameters are used from expert.ai Document Analyzer API to make a decision while marking a conversation as a potential threat to the child. alt text

  4. Alert: If it detects any signs of harmful content, our app will send real-time alerts to the parent’s Falcon app with the conversation screenshot, time and context or tags for which the message was flagged.

  5. Action: Parents can take actions based on the threat to educate a child on online safety. Parents also will receive a summary of what type of content the kid is exposed to on the Internet using keywords, emotions, sentiment as the parameters from expert.ai.

How we built it

Following technologies and data were used to build WatchFalcon:

alt text

  1. Native Android [Java] to create a parent and child app. View-Hierarchy Tracing algorithm along with expert.ai NL APIs on Android, to transverse through the content/text being shown on screen and send it for analysis to expert.ai APIs.
  2. Expert.ai APIs: Document Analyzer and Context Analyzer APIs used with expert.ai java client to extract sentiments, main syncons, main phrases, and classification parameters are used from expert.ai Document Analyzer API to make a decision while marking a conversation as a potential threat to the child.
  3. Google Cloud, Firebase, Python to build the backend.
  4. Dataset: We collected the majority of a dataset of cyberbullying and frauds to basic train the model, but majorly we used expert.ai NL API for our detection system.

Challenges we ran into

As we got to know about this hackathon 2 weeks before the deadline, we had limited time to develop the whole app and we feel if we had more time we would have innovated more on WatchFalcon using expert.ai.

Accomplishments that we're proud of

We were able to build an almost production-ready build in a very short time using expert.ai while creating something which will be useful for almost every parent out there.

Now to use it for me, I can't wait to get married, have kids just to use our app to monitor and mentor them. Just kidding ;)

What we learned

We learnt that every day there are more than 20,000 cases of cyberbullying and cyber frauds against children that go unnoticed every day, which are the main cause of depression and suicidal thought for kids. This could have been avoided if parents had monitored and mentored their kids proactively.

Why WatchFalcon?

Even though there are a lot of parental monitoring systems, there exist almost zero platforms to monitor child's conversation and are limited to few social media communication channels which provide APIs, but our app works independently of any platform or API as we use the view-hirerchary transverse algorithm on android, making our app readily scalable for any medium.

While providing the above specs to the users we make sure the child's privacy is not invaded by parents and help to build a great relationship with them.

As Expert.ai APIs act as a brain of our application and helps us to decide the signs of danger readily while providing massive analysis capabilities, helps us to expand our platform beyond content monitoring.

Everyday 40000 cyberbullying, sexual exploitation and blackmailing happen throughout the globe leading to depression and other mental issues for the child which might impact the moral value in this world. But with WatchFalcon, parents can easily cut off the bad at its root while saving time. Isn't WatchFalcon a way to good parenting?

What's next for watchfalcon,

The Plan: To build an auto-advisory system for kids based on the context of the threat and child' psychology so that kids will know what is best for them without the interference of parents. In that way, we will be able to create a bullying-free world.

Technical Reference: How NLP could be used to detect cyberbullying and cyber frauds? https://www.researchgate.net/publication/310768726_Automatic_Detection_of_Cyberbullying_on_Social_Networks_based_on_Bullying_Features

Built With

Share this project:

Updates