Update README.md

README.md

@@ -1,8 +1,8 @@
 ---
 license: other
-license_name: commoncrawl-derived
+license_name: commoncrawl-tou
 license_link: https://commoncrawl.org/terms-of-use
-pretty_name: Common Crawl Host-Graph Embeddings (deg≥8, shallow LP, dim-128)
+pretty_name: Common Crawl Web Graph Embeddings
 tags:
 - common-crawl
 - web-graph
@@ -20,19 +20,20 @@ configs:
     path: vectors/emb-*.parquet
 ---
 
-# Common Crawl Host-Graph Embeddings — `cc-main-2025-26-nov-dec-jan` (deg≥8, shallow, dim-128)
+# Web Graph Embedings from Common Crawl's Host-level Hyperlink Graph
 
 Dense 128-dimensional embeddings for **52,913,544 web hosts**, learned by link prediction on the
-Common Crawl **host-level hyperlink graph** release `cc-main-2025-26-nov-dec-jan`
-(279.4 M hosts / 13.4 B directed edges). Vectors are **L2-normalized** and served in **float16**;
-similarity is **cosine** (a dot product on the unit vectors).
+Common Crawl **host-level hyperlink graph** release `cc-main-2025-26-nov-dec-jan`. Vectors are **L2-normalized** and served in **float16**;
+similarity is **cosine** (a dot product on the unit vectors). The corresponding source code is available on [Github](https://github.com/commoncrawl/web-graph-embeddings).
+
+> **Scope.** These are the **link degree >= 8** hosts (total in+out degree), a 52.9 M / ~19 % induced subgraph that carries ~97 % of the edges.
+
+![graph-embeddings-visual](https://cdn-uploads.huggingface.co/production/uploads/5efda656ff69163f6f59e5d2/NT89oLi9eBr_3Z-DYfjCr.png)
 
-> **Scope.** These are the **degree ≥ 8** hosts (total in+out degree), a 52.9 M / ~19 % induced
-> subgraph that carries ~97 % of the edges. The low-degree tail of the full 279 M-host graph is
-> **not** in this table — it is intended to be served separately via a domain/metadata fallback.
 
 ## What each row is
 
+
 One host, keyed by two stable string identifiers derived deterministically from the Common Crawl
 host name (no per-release integer ids — those are ephemeral):
 
@@ -61,7 +62,7 @@ Stream as a 🤗 `datasets` table:
 
 ```python
 from datasets import load_dataset
-ds = load_dataset("malteos/cc-web-graph-embeddings", split="train", streaming=True)
+ds = load_dataset("commoncrawl/web-graph-embeddings", split="train", streaming=True)
 row = next(iter(ds))
 print(row["host_key"], len(row["embedding"]))
 ```
@@ -83,26 +84,98 @@ X = np.vstack(mats)                      # [52_913_544, 128], already L2-normali
 - **Source graph:** Common Crawl host webgraph `cc-main-2025-26-nov-dec-jan`, aggregating crawls
   **CC-MAIN-2025-47, CC-MAIN-2025-51, CC-MAIN-2026-04**.
 - **Encoder:** shallow (per-node) embedding table trained by link prediction (WholeGraph-sharded,
-  in-batch negatives), `dim=128`, 3 epochs. `model_space_version = cc_deg8_shallow_v1`.
+  in-batch negatives), `dim=128`, 3 epochs.
 - **Link-prediction quality (held-out edges):** MRR **0.957**, Hits@1 **0.939**, Hits@10 **0.984**.
-- **Precision:** float16 is the **primary published tier**; the lossless float32 master is archived
-  (for L2-normalized unit vectors fp16's cosine error is ~3–4×10⁻⁵ — negligible for ranking).
 
-## What the embedding captures
 
-Downstream probes show the space encodes **graph structure**, not page content-language:
-topic and spam signals are recoverable from the vectors (topic emb-only logistic-regression
-macro-F1 ≈ 0.39), while **content-language is at the floor** (k-NN ≈ 0.02) — hosts are placed by who
-links to whom, not by the language they publish in. Select on topic/spam utility, not LP MRR alone.
+## Explore it in the browser
+
+Two Hugging Face Spaces let you inspect the released vectors without downloading the full dataset.
+
+![2d-viewer-screenshot](https://cdn-uploads.huggingface.co/production/uploads/5efda656ff69163f6f59e5d2/CNDxVBDe9bg4eTuG0MY9Q.png)
+
+**[2D map](https://huggingface.co/spaces/commoncrawl/web-graph-embeddings)** — an interactive
+projection of all 52.9 M hosts, streamed by tile. Pan and zoom, hover any point for its hostname
+and stats, or jump to one of the top 20k hosts by name. The same layout recolours by link degree,
+topic, content language or quality, which makes it easy to see where the embedding separates a
+property and where it does not — the view above is coloured by language.
+
+**[Nearest-neighbour search](https://huggingface.co/spaces/commoncrawl/web-graph-knn)** — type a
+hostname and get its closest hosts in the full 128-dimensional space rather than in the 2D
+projection. All 52,913,544 hosts are indexed. It is the quickest way to check what the embedding
+considers "similar" for a site you already know well.
+
+
+## Evaluation
+
+The model is trained only to predict links, so every result below is a probe of what link structure
+alone turns out to encode. Held-out link prediction is near-saturated (`test_mrr 0.9574`); the
+interesting question is what transfers.
+
+### Topic — the strongest transfer
+
+A logistic probe on the raw vectors reaches **macro-F1 0.388** over 24 WebOrganizer topics, against
+**0.062** for a majority-topic-per-TLD baseline — about **6×** the identity baseline, on 170,399
+labelled hosts. Link structure makes a site's topic close to linearly separable.
+
+### Language — near-perfect on well-connected hosts
+
+A single 120-class macro-F1 badly understates this, because it averages over many classes with
+almost no held-out support. Stratified by link degree the signal is unambiguous (300,000 held-out
+hosts, cosine kNN):
+
+| host degree | accuracy | neighbour-precision | weighted-F1 | macro-F1 @sup≥20 |
+|---|--:|--:|--:|--:|
+| 8–16 | 0.613 | 0.455 | 0.525 | 0.095 |
+| 128–512 | 0.740 | 0.662 | 0.710 | 0.326 |
+| 512–4096 | 0.921 | 0.892 | 0.913 | 0.596 |
+| ≥ 4096 | **0.980** | **0.970** | **0.976** | **0.713** |
+
+Overall accuracy is 0.673 and weighted-F1 0.612. Mid-resource languages climb from the floor to
+strong as degree rises — Spanish 0.02 → 0.45, Polish 0.03 → 0.95, Indonesian 0.02 → 0.83. **The more
+links a host has, the more precisely the embedding places its language.**
+
+### Spam — an independent and unusually durable signal
+
+The embedding never saw a spam label. Compared with **Anti-TrustRank** (ATR), the classic link-based
+spam algorithm, over 52.6 M candidate hosts:
+
+- **They are not the same signal.** Spearman(embedding, ATR) = **−0.04**.
+- **The embedding is the better clean-vs-spam discriminator.** Against operator-reviewed clean
+  sites, gold-AUROC is **0.937** for the graph-propagated embedding, **0.874** for a plain logistic
+  probe, and 0.865 for ATR.
+- **It detects what ATR cannot.** At a 1% false-positive budget against hard, spam-shaped sites that
+  reviewers had manually cleared, the embedding finds **40%** of known spam where ATR finds ~0%; at a
+  5% budget, **74%** vs ~0%.
+- **It cuts false positives 4.5×.** Using ATR for recall and the embedding to vet its top 1,000 drops
+  legitimate false positives from 0.122 to **0.027**, holding spam precision at 0.967.
+- **It survives a crawl change.** On spam labelled from a *later* crawl, ATR's gold-AUROC falls to
+  0.56 — chance — while the embedding still ranks it (enrichment **173×**, AUROC 0.91).
+
+ATR remains the stronger raw ranker on the easy full-web pool (enrichment@1k 117.8×). The practical
+recipe is both: ATR for recall, the embedding to vet its hits and to catch the spam that
+guilt-by-association misses.
+
+**In one line: link structure alone carries a surprising amount of what a site is** — its topic, its
+language wherever the host is well connected, and a spam signal independent of the classic
+link-based one.
+
+## Coverage and limitations
 
-## Limitations
+- **Only the well-connected core.** Training used hosts with total degree ≥ 8: 52.9 M of the crawl's
+  279.4 M hosts (**19%**), covering 13.07 B of ~13.4 B edges. The 81% long tail — hosts with almost
+  no links — is *not* in this artifact. There is no vector for a host that was not trained.
+- **Transductive.** There is no encoder to apply to a new host; a host absent from the training crawl
+  cannot be embedded without retraining.
+- **Direction-blind**, as above.
+- **One crawl window.** Vectors from different releases are not comparable — the model space is
+  arbitrary up to rotation, so a new release means a new space (tracked by `model_space_version` in
+  the manifest).
+- **Spam metrics are a conservative floor.** The label sets are partial, so unlabelled true spam is
+  scored as a false positive; real precision is better than the numbers above.
 
-- Degree ≥ 8 hosts only (see Scope). No vectors for the isolated/low-degree tail.
-- Vectors are comparable **only within** `model_space_version`; a retrain / new window bumps it.
-- Identity keys are pure functions of the published (punycode) host string — no IDNA decoding.
 
 ## License
 
-Derived from the Common Crawl host webgraph; use is subject to the
-[Common Crawl Terms of Use](https://commoncrawl.org/terms-of-use). The embedding artifact is
-released for research use. *(Set the final license before making this dataset public.)*
+Derived from the Common Crawl host webgraph. Its use is subject to the
+[Common Crawl Terms of Use](https://commoncrawl.org/terms-of-use).