How to

Debunker-Assistant is like having your own personal fact-checker, using a wide range of strategies that human readers naturally adopt when analyzing news. It looks at things like style, meaning, and context to give you a set of scores based on the different elements it extracts from the news article. These insights help readers make smarter choices about what to trust online.

Here's a quick breakdown of how it works:
  • UntrustabilityThis score checks the reputation of the news source and its impact on the news ecosystem. If a source has previously published or shared fake news, it's flagged as less trustworthy. The idea is simple: sources that spread fake news are less reliable.
  • ClickbaitWe all know those sensational headlines designed to grab your attention. This feature looks for those tricks, like over-the-top claims or eye-catching thumbnails, which often lead to misleading or exaggerated information.
  • Informal StyleThis feature spots signs of informal language that don’t fit with the objective tone of quality journalism. Things like too many exclamation marks, contractions, or pronouns like "I" or "we" can hint at subjectivity rather than hard facts.
  • ReadabilityThis measures how easy (or hard) the article is to read. A good news piece shouldn’t be too complex or overly simplified. Our system checks things like sentence length and word choice to make sure the content is appropriate for a broad audience.
  • Affective StyleEmotional tone plays a big role in how we react to news. This score detects the use of emotional language that can subtly influence readers and push subjective opinions. Fake news often leverages emotions to spread faster.
  • Dangerous StyleThis feature flags inflammatory or provocative language, which can stir up harmful beliefs or actions. Misinformation thrives on division, and this score helps identify language that could fuel it.

With Debunker-Assistant, we break down these elements to give you a clearer picture of how reliable the news you’re reading really is—making sure you stay informed, not misled!
know what you’re getting

A few frequently asked questions

What's the difference between Debunker-Assistant and a fact-checker?

While traditional fact-checker websites dive deep into analyzing news and label it as fake or true, Debunker-Assistant gives you an instant snapshot of a news article's characteristics. It helps you spot red flags quickly, making it more of a sidekick than a substitute for fact-checkers. Think of it as your trusty assistant in the battle against misinformation!

Who can use Debunker-Assistant?

Anyone and everyone! Whether you're a concerned citizen, a developer, a journalist, or just someone with a curiosity for current events, Debunker-Assistant is your go-to tool. It's designed for anyone who questions the authenticity of news articles, helping you dive deeper into disinformation and its sneaky ways.

How does Debunker-Assistant compute the indicators?

Debunker-Assistant employs advanced AI models and heuristic strategies to pull out key traits from the headline, content, and source domain of the news. All of this is powered by Natural Language Processing and Network Analysis techniques, inspired by the latest research on detecting fake news. It's like having a digital magnifying glass to sift through the facts!

Which languages are supported by Debunker-Assistant?

Currently, Debunker-Assistant can analyze news in Italian and English. Our development team has trained AI models specifically for these languages (you can check them out on Hugging Face/aequa-tech. But we're not stopping there! We plan to expand to Spanish and French in the near future.

I'm a developer. How can I contribute to the fight against disinformation online?

Join the team at Aequa-Tech and weLaika in our mission! We've set up a GitHub organization to engage developers and freelancers in creating new features for both the front and back ends of Debunker-Assistant. Curious to help out? Find more info here.

I'm a journalist or a citizen. How can I contribute to the fight against disinformation online ?

Your insights are invaluable! We're launching a collaborative evaluation campaign from February to April 2025, and we'd love to gather ideas and feedback from experts and everyday citizens alike. Our goal is to refine Debunker-Assistant into a tool that thoroughly analyzes news, with input from our amazing community.