Release v0.1.2-alpha: Best Cluste Selection for Feature Diversity
This release brings significant improvements to the clustering feature.
When multiple clustering runs are performed (clustering_runs configuration), the system now intelligently selects the "best" clustering outcome.
The "best" is determined by a diversity score, which measures the number of unique predominant moods found across the generated playlists. This ensures that the final set of playlists offers a richer and more varied musical representation based on mood characteristics.
Technical Changes:
Backend (app.py):
- run_clustering_task is now a Celery task with built-in progress and log reporting.
Frontend (index.html):
- Added the new field to set the number of cluster execution to run