Scribd
Upload
80 views

Transaction Hypergraph Models For Pattern Identification in The Bitcoin Blockchain

This document discusses research exploring the use of advanced graph models like directed, weighted hypergraphs to represent transactions in the Bitcoin blockchain in order to identify patterns of illicit activity. The researchers are modeling transactions as nodes with addresses as labels in a transaction graph, gathering common addresses into a bipartite multigraph, and contracting parallel edges to create a transaction hypergraph. They outline representations of the blockchain including descriptive statistics and visualizations. The goal is to identify common situations like change-making and mixing services as hypergraph patterns and exploit high-performance computing and graph invariants to analyze the large transaction networks.

Uploaded by

Haber Heisenberg
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
80 views

Transaction Hypergraph Models For Pattern Identification in The Bitcoin Blockchain

This document discusses research exploring the use of advanced graph models like directed, weighted hypergraphs to represent transactions in the Bitcoin blockchain in order to identify patterns of illicit activity. The researchers are modeling transactions as nodes with addresses as labels in a transaction graph, gathering common addresses into a bipartite multigraph, and contracting parallel edges to create a transaction hypergraph. They outline representations of the blockchain including descriptive statistics and visualizations. The goal is to identify common situations like change-making and mixing services as hypergraph patterns and exploit high-performance computing and graph invariants to analyze the large transaction networks.

Uploaded by

Haber Heisenberg
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Transaction Hypergraph Models for

Pattern Identification in the Bitcoin Blockchain

Cliff Joslyn1,? , Chase Dowling1,2 , Sean Kreyling3 , Stephen Ranshous1,4 ,


Curtis West1 , and Amanda White5
1
Data Sciences and Analytics, PNNL, Seattle, WA
2
Electrical Engineering, University of Washington
3
Global Security and Technology Policy, PNNL, Seattle, WA
4
Computer Science, North Carolina State University
5
Computational Mathematics, PNNL, Richland, WA
This poster reports on the recent efforts of our research group to explore the
potential of advanced graph models, including directed, weighted, hypergraphs,
to represent transactions within the Bitcoin blockchain to identify patterns of
illicit activity. Few groups have attempted whole-graph analysis of the Bitcoin
block chain. Our interest is to do so in the context of the range of mathemati-
cal structures most appropriate for this data object, in order to understand the
potential to identify patterns of money laundering. More generally, we seek meth-
ods to analyze Bitcoin as an exemplar of these burgeoning “social machines”:
distributed, autonomous, cryptographic databases poised to automate a range
of social functions, from trade regulation to contracts and public records. We
exploit multiple mathematical representations of the block chain, including:
Transaction Graph: Modeling transactions within blocks as nodes, the result
is an address-labeled weighted directed graph.
Transaction Bipartate Multigraph: Gathering common addresses into a dis-
tinct node type yields potentially multiple weighted edges between them.
Transaction Hypergraph: Contracting parallel edges then results in a di-
rected hypergraph, with transactions as hyperedges joining multiple ad-
dresses as inputs and outputs.
While additional structures like user graphs are commonly analyzed, they
depend on additional information and/or heuristic methods. In contrast, we seek
mathematical and computational methodologies to identify common situations
such as change-making, aggregation, and mixing services as hypergraph patterns
based on their inherent structure and content.
We delineate each of these mathematical representations with examples, in-
cluding descriptive summary statistics and visualizations. We finally report on
our additional work in exploitation of high-performance computation capabili-
ties to handle graphs of O(100M ) size; the use of integer partition measures to
generalize graph degree measurement; and the use of higher order graph invari-
ants such as graph spectra to identify localization of transaction network pivot
points and central entities.
?
Corresponding author, Pacific Northwest National Laboratory, 1100 Dexter Ave.
#400, Seattle, WA 98109, [email protected]

You might also like