Update readme.md

readme.md

@@ -1,23 +1,41 @@
-# Biomedical Fact-Checking App (Hugging Face Spaces)
+# Biomedical News Fact-Checker & Research Evidence Finder
 
-This app:
-- Extracts key scientific claims from a news article,
-- Finds PubMed abstracts,
-- Checks for support/contradiction/neutral stance using advanced NLI,
-- Summarizes all evidence for each claim in plain English using an open-source LLM (Gemma-3B, fully local).
+This app helps you understand if the scientific claims in health or biomedical news articles are supported by real research.
 
-## How to use
+## What does this app do?
 
-1. **Add your `gemma-3b-it-Q4_K_M.gguf` (or smaller) to the repo root.**
-   - Download from https://huggingface.co/unsloth/gemma-3n-E4B-it-GGUF.
-   - Only Q4_K_M or Q4 quantizations will work on Spaces CPU.
+- **Extracts key scientific claims** from a news article.
+- **Finds the most relevant PubMed biomedical research papers** for each claim.
+- **Checks which sentences in those papers support, contradict, or are neutral** to each claim (color-coded: green=support, red=contradict, gray=neutral).
+- **Gives you a simple, plain-English summary** of the overall evidence for each claim.
 
-2. **Deploy on Hugging Face Spaces** (SDK, CPU runtime).
+## How do I use it?
 
-3. **Enter a news URL, see the extracted claims, evidence, and summary verdicts!**
+1. **Paste the link to any health or biomedical news article** (for example, from Medical News Today or similar sites).
+2. **Click the submit/run button.**
+3. **Wait for the results** (this may take a minute, especially the first time you use the app).
+4. For each main claim in the article, you’ll see:
+   - A short summary verdict in plain language.
+   - A list of color-coded evidence sentences from research papers.
+   - The titles of the most relevant PubMed articles.
 
-**No OpenAI or cloud dependencies. All runs on CPU.**
+## Who is this for?
+
+- Curious readers, students, journalists, researchers, or anyone who wants to check if a news story’s science is backed up by real research.
+
+## What is special about this app?
+
+- **No account, no sign-up, no private info required.**
+- **No paid or cloud AI services:** Everything runs here using open-source AI.
+- **Designed for health/biomedical news, but can work on most science articles.**
+
+## Try it out!
+
+**Sample article you can use:**  
+https://www.medicalnewstoday.com/articles/omicron-what-do-we-know-about-the-stealth-variant
+
+Paste this link in the app to see how it works!
 
 ---
 
-**Note:** First summarization may take 10–30s to load LLM, then faster.
+Questions or feedback? You can find the code on Hugging Face Spaces.