Skip to content

Vigilant Bot is an embedded hardware system containing a 9-DOF IMU which detects and acts on distress signals conveyed through complex hand gestures in real time using deep learning.

Notifications You must be signed in to change notification settings

guptajay/NUS-Hack-Roll-2020

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VigilantBot 🚨

Violence against women and girls does not discriminate by race, religion, culture, class or country. Worldwide, one in three women have experienced either physical and/or sexual violence, and more than 15 million girls aged 15-19 years have experienced rape.

News Bias

Perspective Bot is a Chrome Extension where we can search for the news we see on the internet and get more context about it. Using Machine Learning algorithms like BERT (Bidirectional Encoder Representations from Transformers) and clustering, the bot can show different perspectives about the news with the most hits by pulling tweets from the last one week using the Twitter API.

Perspective Bot

Problem Statement

  • Many sources people consider to be ‘news sources’or 'forums' are actually dominated by analysis and opinion pieces. It is hard for people to determine news as 'news' or 'opinions'.
  • People base judgements on limited information, unaware of other view-points, which results in a lack of perspective about global issues.
  • It is important to reduce the bias that the media introduces as well as our own biases.

Solution

Our team built a pipeline based on the state-of-the-art BERT model with Twitter API to gather all the information available on Twitter about the query searched and automatically cluster them based on different opinions grouped by Country using the Google Maps API to reduce bias by presenting different view-points about the same query.

  • Our team integrated Arduino and MATLAB to get live stream data from a BNO055 Inertial Measurement Unit (IMU) sampling at a frequency of 100 Hz attached to a person's wrist.

  • The incoming data in MATLAB is continuously logged into a csv file which is being simultaneously accessed by

  • The selection process begins by gathering all the relevant tweets from the last one week about the query searched using Twitter API.

  • The tweets extracted are passed in the BERT Natural Language Processing model which converts them into detailed high dimensional vectors. The vectors are then automatically clustered using K-Means++ on the basis of different view-points and opinions to gather more perspective about the news. The best representative tweet from each cluster is selected.

  • The results obtained are grouped by country (obtained from Google Maps API), and shown on a website built using BootStrap where the user can gather more context and perspective as well as do further research and reading of the same.

Screenshots

Click Webpage

Installation

  1. Download Google Chrome Extention from here.
  2. In the Chrome Address Bar, type chrome://extensions/ and Toggle Developer Mode to ON (top-right pane)
  3. Using the Load unpacked button, select the folder downloaded in Step 1 to load the app in Chrome and enable the extension.

Team

  1. Rahul Bhattacharjee
  2. Jay Gupta
  3. Mario Josepha
  4. Jason Chow

Built for iNTUition 6.0

About

Vigilant Bot is an embedded hardware system containing a 9-DOF IMU which detects and acts on distress signals conveyed through complex hand gestures in real time using deep learning.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published