the brief
The client proposed that we build a “fake news detector” which would help assist their analysts in market research.
*due to the nature of work, sometimes it is necessary to hide key information about clients
← about this screen
View of an article detail page where an analyst uses aggregated and processed information to aid their work.
the approach
understand
I had to start with some tough questions:
- Does such a thing as a fake news detector exist?
- If not, why has no one solved this before?
You may or may not be surprised to learn that, no, there is no such thing as a “fake news detector”. It has been talked about plenty, and there is a lot of information about fake news, the problems we face with it, and various teams working to create a detector.
Early conversations with the client exposed very useful information, and from these discussions I started to create the scope of the project, gather requirements, build user stories, plan workflows, etc.
← about this screen
A Project screen showing a list of added articles and project notes.
ideate
Once the client and I uncovered the core problems they were facing, we could start discussing how to approach a solution. We worked together to come up with a compromise between the improbable full-fledged fake news detector and a product that still helps to solve their problems.
Throughout this time, it was important to deep dive into research and discovery to define the parameters. I started researching current attempts to create this “detector”, thinking through what output can be expected from a fake news detector, and what could be done to get us as close an approximation as possible.
I mapped out structures for the algorithms the developers would need to build. We worked closely together to check for feasibility as well as ways to improve any of the processes.
Throughout this project we had to continually consider both the systems of things (“how” things work) as well as the problems being faced (the “why” behind what we’re working on).
← about this screen
A public facing article with a public analysis. Releasing public facing analysis of articles were planned for later phases.
design
With a plan of action in place on how to deliver value, and the requirements and user stories mapped out, design got underway.
- Admin & User Management (with multiple user types and their permissions)
- Article & Project Management
- Version control for old & updated articles
- Analyst app
- Public/User facing site
- Plenty more
Designs included:
← about this screen
Public facing "Front Page". Not all articles here would have a full analysis, but a preliminary rating given without review by an analyst. Releasing public facing analysis of articles were planned for later phases.
the result
Starting internally first, the site allowed for a steady inflow of news articles from various sources. The articles filtered through our processing system for:
- Sentiment analysis
- Headline-stance to body-copy matching
- Source subject bias
- Origin detection (the original source of the article, as articles are often republished)
- Key Phrase parsers (for linking articles from various sources and topics in order to get a better snapshot of all information available for a given topic)
The articles can be added to a Project, which is a specific research topic for an analyst to study. Analysis of articles can then be published for the public view, as the client wanted to start opening it up to public use under a subscription model.
contributions
business analysis • ux design • ui design