fn-detect

• 

  Overview for bot account

Inspiration

We tried to complete JP Morgan's & BlackRock's challenges of detecting fake accounts/fake news on social media and creating something positive for the world.

What it does

Chrome extension to add context menu buttons to scan twitter accounts from within twitter, as well as providing tools to perform multi-language lexical analysis on news articles.

How we built it

Python! It's built with:

• Numpy/Scipy
• TextBlob
• SKLearn
• Flask/Python
• Twitter/Botometer APIs
• Yandex Translation API

Machine learning datasets trained on:

• Data from Buzzfeed & Politifact, and open-source datasets from Datacamp
• GloVe vocabulary embeddings from Stanford

Challenges we ran into

Ran into time constraints integrating all components of the toolset into one coherent tool.

Accomplishments that we're proud of

Lexical analysis is really cool, as well as having a chrome extension/functioning website. >90% lexical analysis accuracy with non-linear SVM.

Table Number

15

Built With

• css
• html
• javascript
• jupyter-notebook
• makefile
• python
• shell