Don't Panic!

View by country
View by City
System Overview
Machine Learning Pipeline

Inspiration

Aren't we all tired of hearing day-in and day-out negative news? We decided to break the cycle and bring back positivity by showing that people are full are good sentiment!

What it does

Our service gathers and analyzes the sentiment from extracted news articles from the Schweizer Mediendatenbank (SMD) and a list of Tweets pre-selected by SRF. After being interpreted by our NLP algorithm, everything is displayed on a map by country or by city to compare the general sentiment between news and tweets in different region. You can also filter the news by topics

How we built it

Our service is comprised of two mains components running with two independent life cycles. The first part having the task to process, analyze and extract the sentiment of the articles and the tweets using Natural Language Processing (NLP) provided by the Text Analytics Cognitive Services from Microsoft Azure which will then be uploaded into MongoDB. The second part is a WebApp built with a Spring backend and a Vue.js frontend able to querying and aggregate the data to plot the general sentiment of the news and tweets on a Map built with Google Charts.

Challenges we ran into

Find a meaningful and innovative way to represent the sentiment between two types of media (article vs tweet)
Figure out the impact of the newspaper at a specific location

Accomplishments that we are proud of

Bringing a fully working prototype while successfully integrating multitude of technologies in such a short time.

What we learned

Microsoft Azure Text Analytics Cognitive Services
Google GeoLocation Service with Google Charts

What's next for Don't Panic

Include new news outlets in other countries and continents
Improve the visualization of sentiment over time
Visualize the impact of the media on the tweets

Built With

azure
cognitives
google
java
mongodb
python
spring
textanalytics
vue.js

Submitted to

HackZurich 2020

Created by

I worked on the back-end using Spring framework and also on the Front-end.
I used for the first time Vuejs but knowing React and Angular helped me to easily catch up with Vue.

Mauricio Enriquez Calle
I worked on the NLP part of the project, specifically on developing a sentiment analyzing function that would predict the overall positiveness of the articles and tweets. I also worked on geocoding the datasets. It was my first time using Microsoft's Azure services and Google's Maps API. It was all done in Python. I learned a lot about Python and NLP but also about backened and frontend stuff from my teammates.

Danilo Djordjevic
Student looking to solve challenges employing machine learning, signal processing and statistics
Marcos Carbonell
Venezuelan Computer Science Student at the University of Heidelberg
Thomas Denoréaz