VGo

We are now trying to provide users censored news by detecting fake news and blocking unauthenticated sites, creating an efficient environment for language learners and more.

Comment 
require 'redcarpet'
markdown = Redcarpet.new("VGo is here to protect you from fake news")
puts markdown.to_html

Inspiration

Most students are naive to fake news, while fake news often spread out vigorously. Moreover, students who ought to spend lots of time online studying, for instance, reading online magazines and books to enhance comprehensive reading skills, will be harmed by these fakes news. On the other hand, as more fake news pop up, people need it more to find trustworthy sites. To ultimate news resources, we build a website that will provide our users with the best methods to find and use proper news. It will help students a lot with their studies.

What it does

We are now trying to provide users censored news by detecting fake news and blocking unauthenticated sites.

Fake news will be detected from users' provided URLs. Furthermore, we VGo also provide a grammar check-up service as well as fixing paragraphs and providing a recommendation wording system in Vietnamese. There will be detecting plagiarism for business/school/university account. In the future, we will establish the detecting the spread and popularity of news in order to stop fake news from the danger sites. While reading news, students accessing our website can also practice their reading, vocabulary, and language comprehension skills through practice tests which are compiled by the latest hot news. Look-up dictionary option directly from the passage is available... Another function is the Car News mode, which will convert news Urls to audio so that drivers can also catch up with the lastest censored news while driving. And of course, all news is censored.

How I built it

For real fake news detector: there are two available simple approaches; which are simple classifiers using adam optimizer and binary cross-entropy loss, the other uses long short term memory with the same metrics. The two images below shown the accuracy of these models trained on a small dataset (will be developed later if we have time); which has already yielded an acceptable result for common usage. src/img/realfake_eng_acc.png src/img/models/realfake_lstm_eng_acc.png

For every input URL, we first check whether the URL exists and crawl the information from the article, normalize the information, encode it into a vector, and detect fake news core with the model trained above.

Software development process: Our team will develop this software using a spiral model, we will first start with the most down-on-earth part of this project and will going to advance this app, it will depend on users' reactions...

$$: In-web purchases from a single user, group user (schools, study groups, university), ad banners (free/trial users), the magazines that we recommend (if our project is big enough), donators, etc.

This website will be available for dual languages (Vietnamese and English); though, the Vietnamese part is still under development

Challenges I ran into

Too many things to deal with in more than a day.

Accomplishments that I'm proud of

What I learned

We use GitHub and GitHub page to host our website. This website is built based on HTML, CSS, js, etc. as front end and python on the server-side. We were the one who just wrote "Hello world" yesterday then build a website today. It's kinda crazy, isn't it?

What's next for VGo

An Add on display and

Share this project:

Updates