updated readme

README.md

@@ -17,6 +17,14 @@ size_categories:
 
 # Reddit Tunisia Tech QA Dataset
 
+The pipeline employs a multi-layer scraping strategy to collect Tunisian Arabic tech QA pairs from Reddit.
+It combines three data sources: the Reddit JSON API for real-time posts and comment threads, RSS feeds for post discovery with JSON
+comment fetching, and Arctic Shift for historical data. 
+The system builds both direct QA pairs (post → top comment) and conversation chains (comment → reply) to capture nuanced technical discussions.
+Each pair is scored using a quality function that considers length, tech relevance, and upvotes, then filtered and ranked 
+to produce the top 5,000 high-quality instruction pairs. The final dataset is formatted in Alpaca standard and uploaded 
+to Hugging Face with automatic train/test splitting.
+**results**:
 **5,500 instruction-following QA pairs** scraped from Reddit communities with
 strong Tunisian presence, filtered and ranked for technical relevance.
 
@@ -73,10 +81,10 @@ print(ds["train"][0])
 ## Citation
 
 ```bibtex
-@dataset{tunisian_tech_qa_2025,
+@dataset{tunisian_tech_qa_2026,
   title={Reddit Tunisia Tech QA},
   author={TunisIA-Co-Lab},
-  year={2025},
+  year={2026},
   url={https://huggingface.co/datasets/Farah21/Tunisia_Tech}
 }
 ```