Technology


Newslinn is a technology company focusing on news communication.

As part of our research projects we focus strongly on technical implementation, learnings and external collaboration.

Below is an executive overview of our existing technology. If you have comments or questions please see our blog post on our UGC technology or our LinkedIn Group for more detailed and informal discussions.

Join the conversation

If you wish to participate please join our open Slack channel #newslinn or you can if you prefer.

Expert System

We are using an artificial intelligence system at the heart of Newslinn for validating users and photos. This is using expert driven rules to provide validation and fraud detection based at the granular image level and user/photo level.

Python PyKE Image Data Extraction

Two-Factor Authentication

Built into all workflows is two-factor mobile user authentication, this combined with our expert system forms the basis of our image fraud detection system.

Python Mobile Pin SMS Regional Detection GPS Evaluation

Mobile web app

Built in Framework7 our web app is easily accessible for people and doesn't require people to install it (it's not a smartphone app, but it looks and feels like one). This helps people easily share newsworthy photos with multiple journalists.

Framework7 Javascript Django REST Framework

Real-time Search Engine

Our storage engine is a mixture of cloud base image storage and real-time image indexing system. By combining our real-time expert system with web sockets, we are able to provide near second processing and indexing of images.

MongoDB Socket.io SailsJs Node.js Amazon Cloud Storage (s3)

Email Workflow Validation

We enable users to share photos via vanilla email. This requires incorporating user authentication into an email workflow as well as GPS detection when not enabled on the smartphone.

Python Django SMTP Redis

Photo Feedback & Analytics

We track what happens with photos and what interactions take place within our system and combine that with social media tracking. We also amalgamate journalist feedback and questionnaire data on each photo's usage.

Javascript Django REST Framework nws.li MongoDB

Image Data Extraction

All photos require a data extraction processes before being able to be processed by our expert system. This process extracts metadata along with visual analysis and fraud analysis data. Combining this with our continuing research in supervised photo classification forms the basis of our image fraud detection system. Part of our image data extraction system is combining domain knowledge on the nuances of OS image file saving and software editing patterns.

Python EXIFRead ELA Histogram Analysis