v0.2.2 – Improved ranking algorithm

What's Changed

🔧 Fixed: Ranking algorithm completely rewritten

The file selector (selector.py) has been rewritten from scratch with a much smarter ranking algorithm:

• Multilingual stopword filtering – English, Italian, and code-noise stopwords are filtered out so common words like "con", "the", "function" no longer dominate scoring
• TF-IDF scoring – Term frequency is weighted by inverse document frequency computed across the entire repository
• Path/filename boosting – Files whose path contains query terms get a 3x score boost
• Phrase matching – Bigram and trigram phrases (e.g. "lambda function") are detected and heavily rewarded
• Document length normalization – TF is normalized by document size to prevent large files from unfairly ranking higher
• Coverage bonus – Files matching more distinct query terms get a quadratic bonus

📊 Before vs After

Query: lambda function con logica principale

• Before: ❌ utility_test/test_file_conv.py
• After: ✅ src/lambda_function.py

Query: chatbot main logic

• Before: ❌ irrelevant file
• After: ✅ src/bosa_chatbot/lambda_function.py

Query: lambda function converse API

• Before: ❌ irrelevant file
• After: ✅ src/deep_test_lambda/lambda_function.py

Full Changelog: v0.2.1...v0.2.2

Full Changelog: v0.2.1...v0.2.2

v0.2.1

Full Changelog: v0.1.0...v0.2.1

v0.1.0

Full Changelog: https://github.com/francomano/coreball/commits/v0.1.0