Network Measures of the United States Code
Alexander Lyte
The MITRE Corporation
7515 Colshire Dr.
McLean, VA
alyte@mitre.org
David Slater
The MITRE Corporation
7515 Colshire Dr.
McLean, VA
dslater@mitre.org
Shaun Michel
The MITRE Corporation
7515 Colshire Dr.
McLean, VA
smichel@mitre.org
ABSTRACT
The US Code represents the codification of the laws of the
United States. While it is a well-organized and curated cor-
pus of documents, the legal text remains nearly impenetra-
ble for non-lawyers. In this paper, we treat the US Code as
a citation network and explore its complexity using tradi-
tional network metrics. We find interesting topical patterns
emerge from the citation structure, and begin to interpret
network metrics in the context of the legal corpus. This ap-
proach has potential for determining policy dependency and
robustness, as well as modeling of future policies.
Categories and Subject Descriptors
D.3.3 [Programming Languages]: Python, Neo4j, Cypher,
Javascript; graph theory
General Terms
Graph Theory; Legal Analysis
Keywords
Policy networks, rulesets, United States Code
1. INTRODUCTION
The US Code (USC) is a large, complex, interconnected
corpus of laws that regulate much of American life. With
laws regulating the Armed Forces, Conservation, Banking,
and much more, it not only is an interesting dataset from
a semantic perspective, but also has the potential to reveal
interesting aspects about the US legal regulatory space.
In this paper, we treat the USC as a citation network,
and analyze it using traditional network approaches. We
explore some key phenomenon, including the density of con-
nections, the interrelations among titles, and the emergence
of community structures within the graph.
In section 2, we review previous work on parsing laws,
analyzing their text content, and building citation networks
of interrelated legal documents. Section 3 gives an overview
of the USC, its generation process, organizational structure,
publicly available forms, and an overview of how we con-
struct our citation network. Section 4 walks though base-
line metrics on the graph, including number of nodes, edges,
Approved for Public Release; Distribution Unlimited.
Case Number 15-1489
Copyright 2015 The MITRE Corporation. ALL RIGHTS RESERVED.
degree, betweenness, and centrality by title. Section 5 ex-
plores the interdependencies among the titles, and section
6 describes the results of community detection testing on
the graph. Section 7 concludes with a discussion of future
directions.
2. RELATED WORK
Building citations of legal text is not new. Koniaris out-
lines some of the research done on various legal corpora and
takes a computational approach to parsing the legal text,
defining some standard reference types such as ”amended
by”, ”legal basis”, and ”instruments cited”[4]. The article
provides a framework for integrating various document cor-
pora, such as treaties, legislation, and jurisprudence, and ex-
plores some subgraphs of European Union legislation. Sev-
eral network metrics are established, including degree dis-
tribution, node-edge ratios, and resiliency.
Katz & Bommarito explore the USC from the perspective
of knowledge acquisition, ”a field at the intersection of psy-
chology and computer science”[3]. In doing so, they provide
metrics on the structure, linguistic content, and interdepen-
dence of the USC Titles.
In ”Towards Automated International Law Compliance
Monitoring”, Morgenstern explores the feasibility of parsing
text at the sentence level for use in a rule template frame-
work[5]. Her work develops an architecture for bulk process-
ing of legal text, but notes the challenges in parsing bulleted
text in the Irrealis mood. In parsing the text, Morgenstern
notes that there are different classes of citations, breaking
them down in several ways. First, definitions are treated as
their own type of citation. These are used to build an ontol-
ogy of terms. Second, regulatory citations are classified as
either cross-document, intra-document, or branch, depend-
ing on where the cited document lives. Third, exemptions
are classified in a way that allows for formalization. Lastly,
regulation types are identified as ”obligations, permissions,
prohibitions, penalties, and reparations.”. These classifica-
tions help to codify the law into a set of business rules and
processes.
This recent research helps to clarify the network struc-
ture and content, and provides a framework for mapping
the functions of the law. With these insights and guiding
metrics, we further explore how the function of the law can
be understood by its network structure
3. THE UNITED STATES CODE
The USC represents the compiled federal statutory law of
the United States. It is published every six years by the Of-
