08 JanusGraph Schema for Bitcoin

Bitcoin Graph Relationship Table

信息

Output → Transaction → Output instead of the traditional Input → Transaction → Output

This means removing explicit Input vertices and instead modeling value flow directly through outputs and transactions. In this model:

• An Output is spent in a Transaction (i.e., it's an input to that transaction).
• That Transaction generates new Outputs.
• So the value flow is: Output (spent) → Transaction (that spends it) → Output (created)

This keeps the graph minimal and UTXO-centric — ideal for path analysis and visual tracing. Refined Bitcoin Graph Relationship Table (No Input Node)

From → To	Relationship	Explanation
Output → Transaction	spent_in	The output (used as input) is spent in a transaction.
Transaction → Output	lock_to	The transaction locks value to newly created outputs.
Output → Address	pay_to	The output pays to an address; value is locked to a public key script.
Block → Coinbase Tx	coinbase	The block includes the coinbase transaction, which generates mining rewards.
Transaction → Block	included_in	The transaction included in a block;
Block → Previous Block	chain_to	The block chains to its previous block via prev_block_hash.

Evaluation of This Model:

✅ Strengths

• Simplified graph: Fewer vertex types (no Input nodes), easier to query and visualize.
• Direct value flow: You can trace a satoshi's path from output → tx → output, etc.
• Efficient for UTXO tracing, coin flow, taint analysis, etc.

⚠️ Tradeoffs

• Loss of input metadata: You won’t easily store:

  • input scriptSig / witness
  • input index
  • sequence
  • exact link to the input index of the transaction

• Cannot distinguish input order directly unless it's added as an edge property (spent_in with input_index).

🔧 Recommendations

If you choose this model:

Design Element	Recommendation
spent_in edge	Add properties like input_index, sequence, scriptSig, etc.
lock_to edge	Add output_index, value, scriptPubKey, etc. as properties.
Avoid Input node	Simplifies model. Just make sure metadata is preserved on edges.
Graph traversal queries	Will be simpler for “value flow” and “who paid whom” chains.

Traversal Example

🔁 Value Flow Pattern in This Model

(Output_A) ──[spent_in]──▶ (TX1) ──[lock_to]──▶ (Output_B)

Which translates as:

Output A was spent in TX1, which created Output B.

Here’s a Gremlin traversal example that traces the path from Address A to Address B — in your graph model where the spending relationship is:

Address → Output → Transaction → Output → Address

This represents the flow:

(Address A)
   ↑ pay_to
(Output A)
   ─[spent_in]→ (Transaction)
   ─[lock_to]→ (Output B)
   ↓ pay_to
(Address B)

---

✅ Gremlin Traversal: Address A → ... → Address B

g.V().has('address', 'A')                  // Start at Address A
  .in('pay_to')                            // Go to Output(s) it received funds to
  .out('spent_in')                         // Follow to the Transaction that spent this Output
  .out('lock_to')                          // Follow to Outputs created by that Transaction
  .out('pay_to')                           // Follow to the Addresses those Outputs pay to
  .has('address', 'B')                     // Check if any of them are Address B
  .path()                                  // Return the full path if exists

---

🔍 Optional Enhancements: To trace multi-hop paths (A → ... → B over many txs):

g.V().has('address', 'A')
  .repeat(
     __.in('pay_to')
       .out('spent_in')
       .out('lock_to')
       .out('pay_to')
  )
  .until(has('address', 'B'))
  .path()

• repeat(...).until(...) enables transitive closure — tracing chains of transactions.
• This will trace from A to B even across many intermediate addresses and transactions.

🧠 Tips

• Be sure that your edges have correct directionality:

  • pay_to: from Output → Address
  • lock_to: from Transaction → Output
  • spent_in: from Output → Transaction

• To include edge metadata like value, output_index, etc., use .by(valueMap()) on .path() or project().

JanusGraph Schema

Here's a JanusGraph schema that matches your Refined Bitcoin Graph Relationship Table (No Input Node) — capturing the flow from Output → Transaction → Output, and modeling key Bitcoin concepts like UTXOs, blocks, addresses, and coinbase transactions.

Vertex Labels

mgmt = graph.openManagement()

// Vertex labels
mgmt.makeVertexLabel('address').make()
mgmt.makeVertexLabel('output').make()
mgmt.makeVertexLabel('transaction').make()
mgmt.makeVertexLabel('block').make()

Edge Labels

// Edge labels (directional, multiple allowed)
mgmt.makeEdgeLabel('pay_to').multiplicity(MULTI).make()        // output → address
mgmt.makeEdgeLabel('spent_in').multiplicity(MULTI).make()      // output → transaction
mgmt.makeEdgeLabel('lock_to').multiplicity(MULTI).make()       // transaction → output
mgmt.makeEdgeLabel('coinbase').multiplicity(MULTI).make()      // block → coinbase transaction
mgmt.makeEdgeLabel('belongs_to').multiplicity(MULTI).make()    // transaction → block
mgmt.makeEdgeLabel('chain_to').multiplicity(ONE2ONE).make()    // block → previous block

Properties for Vertices

// Common ID-like keys
addressKey    = mgmt.makePropertyKey('address').dataType(String.class).make()
txidKey       = mgmt.makePropertyKey('txid').dataType(String.class).make()
outputIndex   = mgmt.makePropertyKey('output_index').dataType(Integer.class).make()
value         = mgmt.makePropertyKey('value').dataType(Double.class).make()
scriptPubKey  = mgmt.makePropertyKey('script_pub_key').dataType(String.class).make()
blockHash     = mgmt.makePropertyKey('block_hash').dataType(String.class).make()
blockHeight   = mgmt.makePropertyKey('block_height').dataType(Long.class).make()
timestamp     = mgmt.makePropertyKey('timestamp').dataType(Long.class).make()

// Assign keys to vertex labels (optional, for indexing)
mgmt.addProperties(mgmt.getVertexLabel("address"), addressKey)
mgmt.addProperties(mgmt.getVertexLabel("transaction"), txidKey, timestamp)
mgmt.addProperties(mgmt.getVertexLabel("output"), value, outputIndex, scriptPubKey)
mgmt.addProperties(mgmt.getVertexLabel("block"), blockHash, blockHeight, timestamp)

Indexes (for efficient traversal)

// Vertex-centric index
mgmt.buildIndex('byAddress', Vertex.class).addKey(addressKey).buildCompositeIndex()
mgmt.buildIndex('byTxid', Vertex.class).addKey(txidKey).buildCompositeIndex()
mgmt.buildIndex('byBlockHash', Vertex.class).addKey(blockHash).buildCompositeIndex()

🧠 Notes

Entity	Properties You May Add
output	output_index, value, scriptPubKey, is_spent
transaction	txid, timestamp, fee, size, version
block	block_hash, block_height, timestamp, nonce
address	address (String), maybe type (P2PKH, P2WPKH, etc.)

✅ Sample Relationship Summary

From → To	Label
output → address	pay_to
output → transaction	spent_in
transaction → output	lock_to
block → transaction	coinbase
transaction → block	belongs_to
block → previous block	chain_to