Interactive 2D visualizations of civic issue embeddings using dimensionality reduction
These interactive plots show civic issues from the Járőkelő platform mapped to a 2D space based on their text content similarity. Points that are close together represent issues with similar content, while distant points represent very different types of issues.
Each visualization colors the points by a different metadata field to help identify patterns and clusters in the data.
Technical Details: Embeddings generated using sentence-transformers, dimensionality reduction via t-SNE, visualized with Plotly.
Generated automatically by the Járőkelő Tracker embeddings visualization pipeline.