Document embed + reduce stages in dataset card

README.md

@@ -364,6 +364,91 @@ for sec in structure.get("sections", []):
     print(sec["kind"], sec["label"])
 ```
 
+## Companion stages · `embed` + `reduce`
+
+Alongside the default `documents-*.parquet` shards (one row per
+document, with text + structure), the repo also carries the
+**embed** and **reduce** pipeline outputs as separate parquet
+bundles. Both join back to the `documents` table on the
+`doc_name` primary key.
+
+### `embed-*.parquet` — 2048-D dense vectors
+
+15 shards (~93 MB each, ~1.33 GB total, 10 000 rows per shard,
+deterministic `doc_name` ordering). One row per embeddable
+document (**147,317** rows after dropping the 11 505 NULL-markdown
+rows for which there is no body to embed). Schema mirrors the
+`anle.toaan.gov.vn` corpus's embed stage exactly so cross-corpus
+joins are straightforward:
+
+| Field | Type | Description |
+|---|---|---|
+| `doc_name` | string | Join key back to `documents-*.parquet`. |
+| `text_hash` | string | SHA-256 of the post-normalisation `markdown` (stable across re-runs). |
+| `embedding` | list<float64> | **2 048-D** dense vector from `nvidia/llama-nemotron-embed-1b-v2` (default; other models give other dims). |
+| `embedding_dim` | int64 | Length of `embedding` (denormalised for fast filtering, always `2048` in this release). |
+| `embedding_model_id` | string | Model slug as the embedder backend reports it. |
+| `embedding_text_hash` | string | SHA-256 of the exact text fed to the embedder (differs from `text_hash` when sliding-window chunking applies). |
+| `embedding_chunks_used` | int64 | Number of windows mean-pooled into the final vector (1 when the doc fits in one window). |
+| `embedding_chunking` | string | Chunking strategy: `off` / `sliding` / `sentence`. |
+
+### `reduce-*.parquet` — 2-D projections + cluster ids
+
+15 shards (~0.5 MB each, ~7 MB total, 10 000 rows per shard).
+One row per embeddable document (**147,317** rows). PCA + t-SNE +
+UMAP run with `cfg.reducer.n_components=2` so the `*_z` columns
+that existed in the on-disk per-doc shards are dropped here.
+HDBSCAN cluster ids land in `[-1, 289]` on this corpus (`-1` is
+the noise bucket).
+
+| Field | Type | Description |
+|---|---|---|
+| `doc_name` / `text_hash` | string | Join keys back to `documents-*.parquet` and `embed-*.parquet`. |
+| `pca_x` / `pca_y` | float64 | 2-D PCA projection of the 2048-D embedding. |
+| `tsne_x` / `tsne_y` | float64 | 2-D t-SNE projection. |
+| `umap_x` / `umap_y` | float64 | 2-D UMAP projection (the one used in the scatter PNGs above). |
+| `cluster_id` | int64 | HDBSCAN cluster label; `-1` is the noise / unclustered bucket. |
+
+Quick load (each stage is a `data_files` glob; the default
+`load_dataset("tmquan/vbpl-vn")` still resolves to the
+`documents` config):
+
+```python
+from datasets import load_dataset
+
+embed = load_dataset(
+    "tmquan/vbpl-vn",
+    data_files="embed-*.parquet",
+    split="train",
+)
+print(embed[0]["doc_name"], len(embed[0]["embedding"]))
+# e.g. "100000 2048"
+
+reduce = load_dataset(
+    "tmquan/vbpl-vn",
+    data_files="reduce-*.parquet",
+    split="train",
+)
+print(reduce[0]["doc_name"], reduce[0]["umap_x"], reduce[0]["cluster_id"])
+# e.g. "100000 1.7142 -1"
+```
+
+To join embed-stage vectors back to the document metadata, do
+the join client-side on `doc_name`:
+
+```python
+import pandas as pd
+
+docs   = load_dataset("tmquan/vbpl-vn", split="train").to_pandas()
+embed  = load_dataset("tmquan/vbpl-vn",
+                      data_files="embed-*.parquet",
+                      split="train").to_pandas()
+joined = docs.merge(embed, on="doc_name", how="inner")
+# joined now has the title / so_hieu / markdown / ... columns
+# next to the 2048-d embedding vector for the 147 317 embeddable
+# documents.
+```
+
 ## Trực quan hoá embedding · Embedding visualization
 
 Mỗi điểm là một văn bản pháp luật; toạ độ là vector embedding 2048-D từ `nvidia/llama-nemotron-embed-1b-v2` chiếu xuống 2D bằng UMAP, cụm bằng HDBSCAN. Sáu mặt phân hoạch: `scope`, `doc_type` (mã ngắn), `legal_type` (tên đầy đủ), `legal_area` (lĩnh vực pháp luật), `year`, `cluster_id`. Các nhãn tail-end (sau top-18) được dồn vào nhóm *Khác / Other* màu xám để chú giải đọc được. — Each dot is one legal document; coordinates are the 2D UMAP projection of a 2048-D embedding from `nvidia/llama-nemotron-embed-1b-v2`, with HDBSCAN cluster ids. Six facets: `scope`, `doc_type` (canonical short code), `legal_type` (canonical full name), `legal_area` (subject domain), `year`, `cluster_id`. Tail-end labels beyond the top 18 are collapsed into a grey *Khác / Other* bucket to keep the legend legible. The published reducer parquet still carries `tsne_x` / `tsne_y` columns next to `umap_*` for consumers who want to render t-SNE themselves.