VeriDeJargon [Simplify.ai]

Inspiration

In an age overwhelmed by algorithmically amplified misinformation, it’s harder than ever to find out what’s true and understand it. We were frustrated by how often credible research was misinterpreted in the media, or how ordinary readers couldn’t follow important news because of technical language. We wanted to build a tool that empowers anyone to cut through the noise, spot misinformation, and understand complex content without needing a PhD or a fact-checker.

What it does

VeriDeJargon [Simplify.ai] is a multi-stage NLP pipeline that: -Identifies misinformation by verifying factual claims line-by-line -Generates summaries that only include information marked as reliable and simplifies it

How we built it

Modular Python backend (Flask API) NLTK for text processing and splitting textstat for readability scoring requests + BeautifulSoup for web-sourced definitions Hugging Face Transformers for summarization and misinformation detection Pipeline: clean > split > weight > summarize > de-jargonify > output only reliable, simplified content with the splitting weighing summarizing happening in multiple layers

Challenges we ran into

Defining misinformation granularity: It’s one thing to detect fake news; it’s another to evaluate individual lines for truthfulness without full context. Balancing simplification and accuracy: Oversimplifying technical content risks losing important nuance or scientific precision. Avoiding hallucination in summaries from models like T5, especially when simplifying mixed-factuality input.

What we learned

Not all misinformation is obvious even subtle misrepresentation can drastically alter public understanding Jargon is more than just technical terms it often includes ambiguous framing, hidden assumptions, or inflated claims Combining verification with simplification is more effective than doing either alone Model chaining works surprisingly well when errors are managed correctly

What's next for VeriDeJargon [Simplify.ai]

-We hope to make it a browser extension -Improve its misinformation detection capabilities. -Increasing the capacity of text which it can handle and summarize

Built With

• javascript
• python