vjalili commited on
Commit
40b089a
·
verified ·
1 Parent(s): 249e720

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +44 -1
README.md CHANGED
@@ -6,4 +6,47 @@ tags:
6
  - Cryptocurrency
7
  - Bitcoin
8
  pretty_name: EBA
9
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
  - Cryptocurrency
7
  - Bitcoin
8
  pretty_name: EBA
9
+ ---
10
+
11
+ # Overview
12
+
13
+ [EBA](https://github.com/B1AAB/EBA) interfaces with the Bitcoin network to
14
+ create a graph of the full history of on-chain transactions,
15
+ which includes the complete trading details of `>8.72B` BTC.
16
+ This temporal heterogeneous graph consists of `>2.4B` nodes and `>39.7B`
17
+ time-stamped edges spanning more than a decade,
18
+ making it an ideal resource for developing models on Bitcoin
19
+ and a large-scale benchmark for graph neural networks.
20
+
21
+ Please refer to the following paper for details:
22
+
23
+ > [Jalili, Vahid. "The Temporal Graph of Bitcoin Transactions." The Thirty-ninth Annual Conference on Neural Information Processing Systems Datasets and Benchmarks Track.](https://arxiv.org/abs/2510.20028)
24
+
25
+
26
+
27
+ We share the complete ETL pipeline and all the data it generates.
28
+ To simplify working with the pipeline and its resources,
29
+ we have split them into separate repositories:
30
+ - [ETL pipeline](eba.b1aab.ai/docs/bitcoin/etl/overview);
31
+ - [Dataset release](eba.b1aab.ai/releases/data-release/v1);
32
+ - [Sample communities](https://www.kaggle.com/datasets/vjalili/bitcoin-graph-sampled-communities);
33
+ - [_Hello-World_ model](https://github.com/B1AAB/GraphStudio/tree/main/quickstart/script_classification).
34
+
35
+
36
+ # Block Metadata
37
+
38
+ This dataset provides per-block summary statistics for the Bitcoin blockchain,
39
+ covering all blocks up to height `863 000`. The statistics are derived from four sources:
40
+ parsed from block headers, ETL logs, summarized from block content,
41
+ and generated from the chain in a post-processing step.
42
+
43
+
44
+ The goal of these stats is to provide block-level context;
45
+ either used as an independent resource
46
+ (e.g., to forecast trade volume) or to complement other datasets.
47
+ For instance, they can be combined with the Bitcoin Graph or
48
+ off-chain market indicators (like high, low, open, and close prices)
49
+ to enhance forecasting models.
50
+
51
+
52
+ [Dataset and features documentation](eba.b1aab.ai/docs/bitcoin/datasets/stats)