Te panel identifed a framework for providing computer
security and protecting privacy and civil rights. Tis framework
is shown in Figure S.1 and consists of the following types of
protections:
Data protections ensure that legal backing is there for all
data collected for law enforcement purposes; that covert
and undercover operations using social media analysis
similarly have legal backing; and that the collected infor-
mation is protected from both external and insider threats.
Analysis protections similarly ensure that legal backing
is there for all law enforcement analyses, and that analysis
results are protected from external and insider threats.
More broadly, these protections help ensure equitable
justice outcomes by protecting against inaccuracies and
biases.
Action protections ensure that policing practices are not
distorted and that both enforcement and social service
actions are employed consistently and equitably.
Te core business cases and protection framework elements
are outlined in this report.
Refecting both the business cases and the protection
framework, the expert panel identifed a series of needs for
innovation to better support the use of social media and social
network analysis in law enforcement. Tese needs fall across
four themes that defne an innovation agenda (Figure S.2) to
support the appropriate and sustainable use of these tools for
public safety purposes.
Te frst part of our expert panel’s innovation agenda is
to support working with communities to develop policies and
strategies for using social media and social network analysis.
Here, the initial recommendations relate to developing and
disseminating best practices for transparency and collaborative
decisionmaking for employing social media and social network
analysis technologies, as well as collaboratively creating a series
of model policies for employing and securing these types of
analysis.
Te second part of the agenda is technical research on law
enforcement–specifc social media and social network analysis.
Te initial recommendation is to assess the capabilities of cur-
rent tools and methods and how they might be better tailored
to law enforcement, with the frst step of that assessment being
to create and disseminate a market survey of what tools and
methods are found useful by practitioners now and how well
they are working.
Te third part of the agenda is supporting law
enforcement–specifc training on social media and social net-
work analysis. Here, the initial recommendations are to develop
requirements for training and assess gaps between current com-
mercial- and defense-focused training and what is needed for
law enforcement training. Tis implicitly includes studies and
analyses of what tools and methods are working best in support
of law enforcement operations. Training on legal implications
and protections is a short-term need that can be addressed by
developing a model curriculum.
Te fnal part of the panel’s innovation agenda is a help
desk to help law enforcement agencies navigate requests to
social media companies. Te help desk would help agencies
with making process requests more likely to result in data
returns and/or content takedowns that address the needs of
specifc cases; it would also help agencies process and interpret
the data returned from process requests.
RAND
RR2301-S.1
Analysis
Actions
Community
relations
Data
RAND
RR2301-S.2
1.Enable working
with communities
to develop policies
and strategies for
using social media
and social
network analysis
2.Technical
research for law
enforcement-
specific use of
social media and
social network
analysis
3.Support law
enforcement-
specific training
4.Help desk to
assist agencies
with social media
companies
Figure S.1. An Initial Privacy, Security, and Civil
Rights Protection Framework
Figure S.2. The Innovation AgendaINTRODUCTION
Many modern communication and analytic technologies are
becoming mature enough that they are increasingly accessible
to the average law enforcement organization. Responsible access
and analysis of these technologies hold promise for identifying
and halting crime threats, investigating crimes and holding
ofenders responsible, and detecting and responding efectively
to emergencies and hazards (all of which are core objectives of
law enforcement; see Hollywood et al., 2015, pp. 4–6). At the
same time, law enforcement access to and analyses of commu-
nications data raise concerns about, and require protections for,
individual privacy, civil rights, and information security.
Tis report describes an expert panel’s deliberations on two
such key and closely interlinked communications technolo-
gies:
social media analysis
and
social network analysis
. Te panel
brought together both practitioners and researchers with experi-
ence in using these technologies within law enforcement appli-
cations. Figure 1 summarizes these two technologies and their
relationships.
Social media analysis consists of methods and tools to col-
lect and analyze text, photos, video, and other material shared
via social media systems, such as Facebook, Twitter, YouTube,
Instagram, Pinterest, and Snapchat. Social network analysis is
one type of analysis used in analyzing social media data. Social
media is important today, as a communication and interaction
mode for people in general and as both a “venue” and an enabler
for certain types of crime. Law enforcement interaction with
social media and use of social media data is therefore important,
given the need to police in this technological era. However,
social media analysis by law enforcement does raise acute privacy,
security, and civil rights needs, because of the ubiquitous nature
of the technology and because social media is commonly used for
sensitive and private discussions.
Social network analysis is a type of data analysis that investi-
gates social structures as represented by networks (which can also
be called
graphs
). In these networks, each person is a “node” or
“vertex,” and each relationship between pairs of people is a link
(also called an “edge” or “tie”). Figure 2 shows an example social
network diagram.
Social media, given that it refects relationships inherently, is
a key source of data for social network analysis; conversely, social
network analysis is one key type of social media analysis.
Often, the purpose of social network analysis is to iden-
tify the most “important” or “central” node in a network; how
“important” or “central” is defned varies but is usually based on
the number and types of relationships a person has. As examples:
A person who has more links (i.e., known direct
relationships) than others has a high “degree centrality.”
A person who acts in a bridging role, linking diferent
subgroups that would otherwise not be related to each
other, has a high “betweenness centrality.”
A person who has a high number of indirect relationships,
meaning that the person is related to others who in turn
have a high number of direct links, also tends to have
leadership and infuence.
Also of interest is the ability to recognize subgroups or
subcommunities within a larger social network; a law enforce-
ment example would be to help break a criminal social network
down into likely gangs and cliques within gangs. As an example,
one recent paper identifed methods to recognize gangs and likely
members of gangs within a larger social network (Paulo et al.,
2013).
Figure 3 adds examples of nodes with high degree central-
ity and high betweenness centrality, as well as subgroups, to the
network shown in Figure 2.
When conducting analyses like this, a node is typically a
person, but it does not have to be. Social network analysis can
Social
network
analysis
Social
media
analysis
Major application of social
network analysis is to examine
connections in social media data
RAND
RR2301-1
Figure 1. Social Media and Social Network Analysis
SOURCE: Coldren and Markovic, 2015, p. 14.
RAND
RR2301-3
Tie
Node
Figure 2. Example Social Network Diagram
(Sociogram)also be used to identify important links to assets (vehicles, fnan-
cial accounts) and addresses.
Scott (2012) and Hannemann and Riddle (2005) provide
example textbooks that are useful in educating analysts on how
to make use of social network analysis. Ahajjam, El Haddad,
and Badir (2018) provide an example of a contemporary method
for identifying likely infuencers and leaders in social networks
(in brief, these infuencers/leaders typically have higher numbers
of both direct and indirect relationships). Girvan and Newman
(2002) provide an example of a method for fnding likely sub-
groups within larger social networks.
Given that social media data are inherently networked,
providing information about people and their relationships as
symbolized through communications, analysis of social media
data often relies on social network analysis. Te Social Media
Research Foundation’s NodeXL is an example of a social network
analysis tool that includes capabilities to import data from social
media platforms, including Facebook, Twitter, YouTube, and
Flickr (2016). Pew Research Center (2014) provides an example
of conducting social network analysis on Twitter networks. Tat
said, social network analysis can use any law enforcement data
that provide information on entities (people, locations, addresses,
etc.) and their relationships. Examples include record manage-
ment system (RMS) enforcement and incident data, feld reports
and interview data, tips from the public, and call-for-service data.
As the use of social media platforms has become almost
ubiquitous in modern society, including among ofenders and
organized crime networks, social media is becoming a key source
of information about both threatened and actual criminal activ-
ity. Tere have been multiple high-profle cases where, after a
violent act has already been perpetrated, investigators found what
appear to have been indicators or “warning signs” that might
have been detected and followed up on to prevent the event. Te
feld of social network analysis studies the relationships between
people and assets and can, among other things, identify those
with “central” roles in criminal networks; social network analysis
naturally provides methods for analyzing social media data for
investigative purposes.
Recent cases, law enforcement presentations, and published
accounts provide a range of examples of the use of social media
data and social network analysis for law enforcement purposes:
Solving a gang-related shooting by matching knowledge
about feuding gang networks with motor vehicle
information. Analysts used social network information
to identify potential adversaries of a victim with gang
connections. Tey then queried to see which gang associate
owned the vehicle that matched a witness’s description
(Cheung and Prox, 2012).
Watching posts and videos uploaded to YouTube made by
a particular gang, in which gang members described their
criminal activities and made explicit threats against others.
Tis use does require being able to distinguish between
genuine criminal evidence and “false positive” postings and
relationships (e.g., Popper, 2014). More broadly, police have
used evidence of crimes posted online, including ofenders
posting photos of crime scenes and bragging about them,
to hold ofenders accountable (Dughi, 2016).
Prioritizing subjects for gang call-ins and enhanced
enforcement and prosecution based on how central they
were in criminal gang networks (Coldren and Markovic,
2015).
Social media is growing in importance to law enforcement.
LexisNexis has reported that, as of a 2014 survey of law enforce-
ment professionals who use social media operationally at least to
some extent, 86 percent used social media for investigations two
to three times per month, and 25 percent reported using it daily
(LexisNexis Risk Solutions, 2014). At the same time, there have
been substantial concerns about the usage of these tools. In the
same survey, only 48 percent of respondents said that their agen-
cies had formal processes on social media investigations, and only
9 percent reported receiving training from their agency.
With respect to social network analysis, the mapping of net-
works has been a part of law enforcement investigation for many
years, and new tools provide increased capability. Using tools
that generate networking diagrams (also known as “link charts”)
is common, and the creation of networking diagrams is a core
feature in tools such as Coplink Analyst’s Notebook and Palantir
Technologies’ products. However, the spread of more-advanced
algorithms that leverage cutting-edge academic research in social
network analysis is more limited, and those techniques are much
less known in law enforcement (e.g., Coldren and Markovic,
2015).
RAND
RR2301-3
Group 1Group 2
Group 3
(isolated node)High betweenness centrality
High degree centrality
Tie
Node
Figure 3. Central Nodes and Network SubgroupsMETHODOLOGY
To assess the expanding importance of social media and social
network analysis in law enforcement, we assembled an expert
panel to (1) consider applications and protections for employing
social media and social network analysis and (2) then identify
and prioritize needs for innovation related to use of these types
of analysis and associated analytics in law enforcement. Panel
members included a range of experts, including practitioners
of social media and social network analysis in law enforce-
ment, developers of social media and social network analysis
methods for law enforcement, and researchers and attorneys
with expertise on community advocacy, privacy, and civil rights
issues related to these types of analysis. Panelists were identifed
through a collection of literature reviews (including both scien-
tifc articles and recent conference and workshop presentations)
and assistance from National Institute of Justice and Bureau of
Justice Assistance staf.
We engaged the panel to achieve two purposes. Te frst,
given the relative newness of using these types of analysis in
law enforcement, was to provide insight to the law enforcement
community, including practitioners, funders, and developers,
on how these technologies might be used and secured efec-
tively. Tis included
identifying emerging applications for using social media
data and social network analysis in law enforcement, and
capturing key process steps and considerations in business
cases
identifying a core set of security, privacy, and civil rights
protections when using social media data and social
network analysis in law enforcement.
Given that law enforcement use of social media and social
network analysis is still relatively new and is controversial for a
variety of reasons, the intent was to contribute to and advance
policy debate on these issues. Te efort was intended to survey
both the promise and challenges of these technologies, and
to frame areas where additional research and attention were
warranted,
not
to provide defnitive guidance or propose model
policy for their use.
Te second purpose was to identify specifc needs for inno-
vation to help law enforcement make better use of social media
data and social network analysis. We defne a need as a require-
ment put forward by the panel for research, development, or
dissemination of a product or service to help solve a problem
or take advantage of an opportunity. “Products” can include
nonmaterial items such as new policies, regulations, processes,
analytic techniques, and organizational structures, in addition
to physical systems.
Appendix A, the Technical Supplement, provides the
technical details of the generation and prioritization of the
needs. In summary, to frame the panel discussion, prior to the
workshop, we sent out a read-ahead and had participants fll
out an online questionnaire. Te questionnaire asked panelists
to identify specifc operational questions that social media data
Workshop Participants
Jeff Asher
Journalist; formerly with New Orleans Police Department
Charles L. Cohen
Indiana State Police
Dawn Diedrich
Georgia Bureau of Investigation
Andrew Ferguson
University of the District of Columbia
Kevin Hiner
City of Richmond, Va., Police Department
Rachel Levinson-Waldman
Brennan Center
John Markovic
Bureau of Justice Assistance
Michael G. Mastronardy
Ocean County, N.J., Sheriff’s Offce
Joe McHale
Marion, Iowa, Police Department
Patrick Muscat
Wayne County, Michigan, Prosecutor’s Offce
Desmond Patton
Columbia University School of Social Work
Jay Stanley
American Civil Liberties Union
Wendy H. Stiver
Dayton, Ohio, Police Department
Lee Tien
Electronic Frontier Foundation
Michael Yu
Montgomery County, Md., Department of Police
5。