Content uploaded by Dan Suthers
Author content
All content in this area was uploaded by Dan Suthers on Nov 07, 2023
Content may be subject to copyright.
Detection of Domestic Human Trafficking Indicators and Movement Trends
Using Content Available on Open Internet Sources
Michelle Ibanez
Communication and Information Sciences
University of Hawaii
ibanezm@hawaii.edu
Daniel D. Suthers
Information and Computer Sciences
University of Hawaii
suthers@hawaiii.edu
Abstract
This study applies network analysis to domestic
human trafficking activity in online environments to
identify trafficking circuits. Numerous websites host
online classifieds, discussion forums, review sites, and
social media related to sex trafficking. Trafficked
persons are systematically moved to various locations
throughout the U.S. based on demand, and traffickers
use online classified sites to advertise their victims.
The goals of this study were to identify indicators of
sex trafficking in online advertisements posted on open
Internet sources and to derive movement patterns.
Online classified ads for adult services in Hawai‘i
were collected over a six-week time frame and
assessed for indicators of human trafficking. Data
captured in the analysis was used to detect movement
trends of potential trafficked persons and mapped to
visualize domestic circuits. A key element in observing
movement was the advertised phone number, as it is
linked to the individual through online advertisements
and customer reviews.
1 Introduction
Human trafficking is a form of modern day
slavery, which entails illegal trade in persons for forced
labor or sexual exploitation. Although slavery is
illegal, it persists today with numbers twice as high as
the trans-Atlantic slave trade [1]. Human trafficking is
the second most profitable organized crime in the
world with an estimated annual global profit of $31.6
billion [2]. Nearly half of all profits are generated in
industrialized countries such as the United States. The
U.S. is the second leading destination country for the
sexual exploitation of trafficked women and children
[3]. In recent years, there has been an increase in
attention from the media, government, and law
enforcement on international human trafficking issues
[4]. However, less attention has been given to domestic
human trafficking. The Trafficking Victims Protection
Act of 2000 (TVPA) is the first U.S. federal law to
address this problem. The TVPA [5] defines sex
trafficking as “the recruitment, harboring,
transportation, provision, or obtaining of a person for
the purpose of a commercial sex act” and defines
severe forms of trafficking as “sex trafficking in which
a commercial sex act is induced by force, fraud, or
coercion, or in which the person induced to perform
such act has not attained 18 years.”
The influx of information and communication
technologies (ICTs) has contributed to this growing
issue. Technology has been cited as a significant factor
facilitating human trafficking [6, 7, 8, 9]. The Internet
and the use of online classifieds have changed the
environment of sex trafficking. Activities have shifted
from a predominantly physical environment to an
increasingly virtual environment, with trafficked
persons often being advertised online as well as on the
street. Virtual red light districts provide a low risk
environment for buyers to connect with sellers [1, 4, 6,
7]. Trafficking activity has been documented in chat
rooms, social network sites, online classifieds, and
social media sites allowing traffickers to exploit a
greater number of victims [7].
Mobile phones serve as the conduit between the
virtual and physical environments connecting clients to
the product for coordination of services. Phones are
typically prepaid mobiles, so they cannot be linked to a
specific individual through a service contract [10].
However, they can be used to provide other clues to
this activity. The majority of ads include a phone
number to contact the poster and schedule services.
Coordination of scheduling may be done through a
central line, with phones linked to the provider via the
trafficker or a decentralized manner, with providers
maintaining individual numbers [1]. Customer reviews
are often linked to phone numbers, indicating the
importance of maintaining individual phone numbers
for reputation development. Research on the use of
ICTs in facilitating trafficking is emerging, but
significant gaps exist [7].
Ibanez, M., & Suthers, D. D. (2014). Detection of domestic human trafficking indicators and movement trends using content
available on open Internet sources. In Proceedings of the Hawaii International Conference on the System Sciences (HICSS-47),
January 6-9, 2013, Hilton Waikoloa, Hawai‘i (CD-ROM). New Brunswick: Institute of Electrical and Electronics Engineers,
Inc. (IEEE).
This is an exploratory study to better understand
and develop methods for tracking domestic sex
trafficking activity online. Trafficked persons are a
subset of commercial sex workers. The focus of this
study was to identify indicators of domestic sex
trafficking in online advertising for commercial sex in
order to identify potential victims and detect movement
trends. Network analysis was used to analyze content
available on open Internet sources to gain insight on
this covert network. Sparrow [11] argued for the use of
social network analysis (SNA) as an effective method
to exploit criminal data, citing its ability to highlight
network vulnerabilities and transform raw data into
intelligence. SNA techniques can be used to process
large volumes of data to detect hidden structures and
patterns of criminal networks. Klerks [12] pointed out
that SNA has primarily focused on positive networks
and avoided dark networks. It was not until the early
1990s that academics began to apply these skills to
studying covert networks. Using SNA methods to
exploit criminal data has the potential to make covert
activity more visible. The primary research question is
whether content available on open Internet sources can
be used to detect trends in movement of potential
traffickers or victims (i.e. hubs, circuits, etc).
2 Domestic Sex Trafficking
Research has predominantly focused on
transnational issues of trafficking, while limited
research has focused on domestic trafficking of U.S.
citizens. According to current estimates, although
many trafficked persons are foreign nationals, a much
larger number are U.S. citizens being trafficked within
the country [13]. Each year an estimated 100,000 –
300,000 American children under the age of 18 are at
risk of being victims of sexual exploitation, suggesting
that the most prominent group of trafficked victims
within the country are U.S. minors [14]. By targeting
and recruiting children (under the age of 18),
traffickers fuel the supply chain as victims are
introduced and groomed into this lifestyle at an early
age. Hanna [15] found “an estimated eighty percent of
adult prostitutes started working as children. Research
indicates that most adult women who work in the
commercial sex industry started working between the
ages of 14 and 18 - a time when they were far too
young to make an informed decision.”
2.1 Definitions of Terms
This criminal network uses various terms to
describe roles and activity. Three main roles of this
trade involve the trafficker (third party seller), the
product (person being sold), and the client. A pimp or
madam is the term used for the potential trafficker. The
product (person) being sold is referred to as the
provider. The group of providers controlled by a pimp
is called a stable. The clients of this trade are known as
Johns. The John community is referred to as
Hobbyists, Punters, or Mongers. This activity occurs in
various locations. The physical location the provider
walks is known as the track. Hubs are areas known for
high volumes of trafficking activity. The series of cities
a provider is sold in is called a circuit. In this paper we
use the term victims to refer to involuntary or coerced
providers.
2.2 Movement along Circuits
Movement is a critical element of trafficking. It
does not define trafficking, but it is a key indicator of
potential trafficking activity. Providers are shuffled
across the nation to various cities based on demand.
Movement typically follows a pattern, with providers
being moved from city to city on a circuit. A circuit is
the systematic movement of providers between cities,
which could encompass entire regions [16]. This
systematic movement of providers to various cities
across the nation for the purposes of commercial sex
activity (profit) alludes to a more sophisticated,
organized criminal activity taking place. A well-
known circuit is the Western Circuit, which includes
Seattle, Washington; Portland, Oregon; San Francisco,
Los Angeles, and San Diego, California; Hawai‘i;
Phoenix, Arizona; Denver, Colorado; and Salt Lake
City, Utah. This data is derived from national crime
statistics based on counts of arrests in locations [17].
The movement of victims by traffickers is driven
by multiple factors. Traffickers remain in a location for
a limited time and move frequently in an attempt to
avoid detection from law enforcement [17]. Frequent
movement is also a control mechanism used by
traffickers to keep victims isolated. The continuous
movement and short durations of stay prevent victims
from establishing social support systems and limits
familiarity with a location [1]. A significant factor
driving movement is market demand. Movement trends
are profit driven. Rotating providers in and out of
different cities keeps the supply stream fresh. Providers
are advertised as being ‘new’ or ‘available for a limited
time only’ enticing the demand side of the market. Due
to the illicit nature of this activity and limited research
conducted on domestic sex trafficking of U.S. citizens,
the exact number of victims in the U.S. is unknown.
However, human trafficking cases have been reported
in all fifty states [18].
2.3 Significance of Phone Number
The advertised phone number is a significant
element in observing movement for a number of
reasons. The phone number is the link to the provider
enabling the John to make contact and schedule
services. The area code provides information on the
origin of the phone, which may indicate where the
provider or trafficker is from [1]. Phone numbers are
embedded in the online classifieds as providers are
advertised along circuits. These ads can be accessed
using a phone number search, which will indicate other
advertised locations. Advertised phone numbers are
also linked to providers’ reputations and product
branding through customer reviews. Provider review
sites use phone numbers as a search term to access
provider history, reviews, current location, etc. In
advertisements, providers often refer to their phone
number as a search term to be used to access their
reviews and indicate if their number has recently
changed.
2.4 Online Activity
The affordances of technology are changing the
sex trafficking industry, with many of the activities and
practices surrounding sex trafficking moving to the
virtual environment [6]. There are numerous websites
dedicated to this community, such as Myredbook.com,
TheEroticReview.com, CityVibe.com, and
NaughtyReviews.com. Lee and Lee [19] state,
“communities no longer exist only in the physical
world but also in the virtual world that operates
through the Internet.” There is a spectrum of behavior
in the virtual environment based on the purpose of a
site and community needs. According to Lee and Lee
[19], users participate in virtual communities to
“garner mutual benefits between group members, for
example, strengthening social ties, circulating
information, archiving experiences and exchanging
opinions.” Although the behavior of this community
may be deemed deviant, their online activities are
consistent with other communities in terms of virtual
community development and participation.
Criminal network data is deceptive; information
may be intentionally misleading, inaccurate, out-of-
date, and incomplete [11, 12, 20]. This makes studying
covert networks challenging. However, as human
trafficking activity increasingly shifts to the virtual
environment it becomes more visible [6]. The Internet
provides users with a sense of anonymity, which makes
participation in this illicit activity feel more discreet
online than offline. This false sense of security creates
a space for this activity to flourish. The community
uses the Internet as a communication channel
providing access to the product and to other members
of the community. They participate in information
seeking and sharing activities with other community
members in open forums. Using the Internet as their
communication platform creates an archive of
information exchange. The key is to identify which
sites and information are useful in detecting sex
trafficking and how that information can be used to
disrupt this activity.
On the supply side, traffickers provide a product
via the Internet. They use online classifieds to advertise
the sale of women and/or children for the purposes of
sexual exploitation. Traffickers are no longer bound by
geographic limitations. They are able to expand their
scope of activity through the use of the Internet. Online
classifieds provide an optimal means of advertising
that is capable of reaching a large audience regardless
of geographic location. As this trend develops, cases of
sex trafficking have been observed far beyond major
city locales and are reaching into remote locations [1].
Ads are often placed in locations prior to arrival for
pre-scheduling of appointments to ensure travel is
profitable.
On the demand side, Johns use the Internet to
search for providers, share information about
providers, compare experiences, and provide warnings
about potential law enforcement [8, 9]. The Internet
provides a virtual catalog of women with access to an
abundance of advertisements and reviews to assist in
the purchasing of a provider that fits the client’s
preference. Johns are able to browse through provider
profiles in order to gain information about performance
ratings and services provided. This grants Johns
anonymous access to a far greater number of providers
than possible offline. boyd et al. [6] report searching
online allows Johns to “remain invisible to law
enforcement who have not yet developed sophisticated
digital operations.”
3 Methods
3.1 Study Overview
The intent of the study was to examine online
adult service advertisements in order to observe the
types of data available in these ads and to identify
ways to transform this data into meaningful
information that can be used to disrupt potential
criminal activity. Online advertisements for adult
services from Backpage Hawai‘i were collected from
January 09, 2013 through February 12, 2013. Due to ad
volume, ads were collected in one-week increments
every other week during the six-week study period,
capturing three weeks of data. The methodology used
in the analysis involved a three-step process. First, an
audit of Hawai‘i Backpage escort ads was conducted
and analyzed for indicators of trafficking. Next, phone
numbers were extracted from the ads and used to track
movement based on area code origin and other
advertised locations. Lastly, data obtained from step
two was analyzed using social network analysis
methods in order to create a visualization of potential
circuits.
Backpage is an online classified site that hosts
advertisements for a wide range of products, including
adult services, which can be found under the ‘Escort’
section. Backpage was selected for analysis because at
this writing it is the leading U.S. site for advertising
prostitution [21]. “Other websites also offer adult
classifieds, but Backpage not only has the highest
frequency of posts but also greater website traffic (user
hits) than most alternative choices, generating an
estimated $1.95 million in revenue in June of 2011
alone” [1]. The cost of escort ads average $3 to $15 per
posting and $7 to $20 for reposting [21]. Also, because
Backpage is a mainstream online classified site, less
stigma is attached to visiting this site than a John site.
Online classifieds in Hawai‘i were selected
because Hawai‘i is a destination location for
trafficking. McClain and Garrity [22] define
destination cities as areas with the greatest demand,
typically locations near military bases, truck stops,
conventions, and tourist areas. The multiple military
bases, large international conferences, and prime
tourist destinations make Hawai‘i a thriving location
for sex trafficking. Another reason Hawai‘i was
selected was due to its geographic location. A
significant element of this study was to detect likely
trafficking activity and map movement. Movement to
Hawai‘i entails a higher risk due to the security
measures associated with air travel, cost of air travel,
and the inability to earn profits enroute. On the
continental U.S., providers are moved along major
roadways and sold at truck stops along the way,
allowing traffickers to continue operations during
travel [1]. Yet in spite of the costs and risks, Hawai‘i
being a destination location with a thriving market
ensures that travel is profitable.
3.2 Content Analysis
An audit of Hawai‘i Backpage escort
advertisements was conducted during the study period
and analyzed for indicators of online human
trafficking. Ads were analyzed for the presence of
online trafficking indicators using an index developed
by the author after an initial study of Backpage
advertisements (Table 1). The indicators used in the
index were derived from a larger list of sex trafficking
and domestic minor sex trafficking indicators produced
by the United Nations Office on Drugs and Crime [23]
and the Polaris Project [24]. The indicators were
selected based on observability in online
advertisements. The raw data was extracted and
converted into research results based on these
indicators.
Important fields of data include advertised
location, advertised age, advertised name, phone
number, area code origin, transitory language, and
miscellaneous observations (i.e. explicit
language/photos). Similar methods were used in
previous Backpage studies conducted by Operation
Broken Silence [1, 25]. These studies were concerned
with identifying the existence of sex trafficking in
Backpage adult services ads. The current study
expands on previous studies by examining the
movement trends associated with potential trafficking
activity linked to online advertisements. Advertised
location indicates where the provider is operating.
Advertised age is the listed age of the poster (typically
inaccurate with many ads using ages older or younger
than true age). Advertised name serves as an identifier
to distinguish unique posts and providers. Both
advertised age and name fields provide information
relevant to indicators of trafficking if inconsistencies
are observed. Phone numbers were used to distinguish
unique posts and provide data on movement. Area code
origin can indicate movement if the origin is different
Table 1. Online Human Trafficking Indicators
Indicator
Explanation
Different Ages Used
(Inconsistencies in
story)
Discrepancies in age
within or across ads
Different Aliases
Used
(Inconsistencies in
story)
Discrepancies in aliases
within or across ads
Movement
(Frequent movement
to work)
Transient language, out
of state area code, ad
posting in different
locations
Shared Management
(Travel in groups)
Ads reference multiple
providers, shared phone
Ad posted by third
party
Third person language
used in ad
Advertised
Ethnicity/Nationality
Ad includes references
to ethnicity or nationality
Potential Restricted
Movement
Incalls only –provider
may be restricted to hotel
room, massage parlor,
etc.
from the advertised location. Area codes may also
present clues to the source location of providers [1].
Transitory language is any form of language that
would indicate movement or travel (i.e. new in town,
limited time only, just visiting, etc.). A field for
miscellaneous observations was collected for any
information that did not fall into the identified data
fields or indicator index.
3.3 Phone Number Analysis
All unique phone numbers occurring in our data
were analyzed to detect and map movement trends,
except that phone numbers belonging to fixed locations
were excluded (escort services and massage parlors).
Movement was determined based on area code origin
and other advertised locations. The area code provides
information on the origin of the phone. Area code
origin served as an indicator of movement if the
location of origin was different from the current
advertised location (i.e. if the phone number is an out-
of-state number it was in at least two locations, the
source location and currently advertised in Hawai‘i –
data source). Movement was further analyzed using a
phone number analysis across multiple John sites to
identify other advertised locations. John sites allow
clients to validate provider authenticity by reviewing
advertisement history. These sites include information
on the various cities providers have been advertised in,
and different ages, aliases, reviews, and photos used in
the ads. These sites serve the community’s needs by
providing a single source of consolidated information
on a provider, allowing Johns to verify information by
cross-analyzing multiple ads for consistency. This
process serves two purposes: it allows Johns to avoid
false advertisement (bait and switch), and avoid
potential law enforcement operations. The various
locations each phone number was advertised in were
documented allowing movement patterns to be
observed.
3.4 Movement Analysis
Provider location data was analyzed using network
analysis methods to detect and map circuits with GIS
data. A provider by location bipartite network was
constructed to analyze movement trends. The phone
number served as a proxy for the provider or trafficker.
Location data was entered using area code origin and
other advertised locations based on the data captured in
the phone number search. Circuits were calculated
based on aggregate data using a monopartite projection
of the network, and visualized. Specifically, the
network was folded to create a location-to-location
network. Locations sharing telephone numbers were
linked, with the link weighted by number of phone
numbers shared between locations. Then, thresholds
were established to filter data by edge weight during
visualization in order to identify high volume travel
routes. GIS data was incorporated in order to overlay
circuits on a map. The resulting visualization of high
volumen associations between locations was used to
identify prominent hubs and circuits. Analysis was
performed using the *ORA software suite [26]. *ORA
is developed by the Center for Computational Analysis
of Social and Organizational Systems (CASOS) at
Carnegie Mellon to assess and analyze dynamic meta-
networks.
4 Results
During the study period 1881 escort ads were
collected and audited, which is an average of 90 ads
per day over a 21-day sample period. After excluding
daily duplicate ads 1436 ads were analyzed. The
advertised location of each ad was documented to
identify an intra-state circuit. The advertised locations
within the state were spread over four of the islands
(O‘ahu, Hawai‘i, Maui, and Kaua‘i). Total ads by
location include: 1255 (78%) in O‘ahu, 210 (13%) in
Maui, 81 (5%) Hawai‘i Island, and 58 (4%) in Kaua‘i.
Multiple ads (56) listed all four islands as the service
location. Based on the advertised location of the
sample, findings suggested the presence of a micro-
circuit through several of the Hawaiian Islands,
indicating that demand exists and air travel to such
locations is profitable (Figure 1). The observance of
this micro-circuit illustrates the ability for traffickers to
extend their reach beyond prominent cities to remote
locations via online advertisements.
Figure 1. Hawai‘i Micro-circuit
4.1 Indicator Results
The indicator analysis measured the number of
indicators present in an advertisement and the
frequency of an indicator across the sample. Of the
1436 advertisements analyzed, 82% of the ads
contained one or more indicators and 26% of the ads
contained three or more indicators. The number of
indicators observed per ad is presented in Table 2. The
distribution of ads by indicator was: 685 (48%)
Advertised Ethnicity/Nationality, 553 (39%) Potential
Restricted Movement, 473 (33%) Movement along
Circuit, 426 (30%) Multiple Providers, 261 (18%)
Third Party Post, 33 (2%) Different Ages, and 8 (1%)
Different Aliases. The presence of sex trafficking
indicators does not prove trafficking is occurring, but it
raises flags to potential activity that requires further
investigation. This information could be used to
identify high-risk advertisements, narrowing down the
pool of ads warranting law enforcement attention.
Table 2. Total indicators present per ad
Indicators per Ad
Total ads
% Ads
6
26
2%
5
104
7%
4
86
6%
3
155
11%
2
365
25%
1
448
31%
0
252
18%
4.2 Movement Results
A total of 234 unique phone numbers were
recorded during the study period. After excluding fixed
locations (i.e. massage parlors or escort services), 208
phone numbers were analyzed. Of those, 165 phone
numbers indicated movement. The number of
advertised locations ranged from two to 44 cities with
an average of six. A total of 44% of the phone number
area codes were from Hawai‘i and the remaining 66%
were dispersed throughout 23 states. The top five area
code origins outside of Hawai‘i included California,
Nevada, Oregon, New York, and Washington. This
information could be used to identify potential source
locations of traffickers or providers. After analyzing
other advertised locations (advertisement history) links
to all but four states (Delaware, Maine, New
Hampshire, and South Dakota) were observed in the
sample, suggesting that Hawai‘i is a destination hub for
this activity. Figure 2 and 3 below provide information
on hub locations observed in the sample.
Based on the sample, Portland, Oregon is among
the top 10 hub cities with movement trends through
Hawai‘i. This is significant in terms of local current
events. In May 2013, a ‘traveling escort’ from
Portland, Oregon was found murdered on the island of
O‘ahu. The media report indicated that she was
‘visiting’ Hawai‘i and she had posted several online
advertisements marketing her services in Waikiki over
the past year. The last advertisement posted was a joint
ad offering services with a traveling companion. The
details provided in the report are indicative of human
trafficking activity. The article can be found at
http://www.oregonlive.com/pacific-northwest-
news/index.ssf/2013/05/portland_woman_ivy_harris_f
oun.html. Using the methods outlined above her ads
would be flagged as high-risk.
0
10
20
30
40
50
60
70
80
90
100
CA NV OR TX WA FL CO NY AZ DC
Top 10 States Observed
Total Phone
Numbers Linked
to State
Figure 2. Frequency Distribution of Advertised
Phone Numbers by State
0
20
40
60
80
100
120
140
160
180
200
Honolulu, HI
Maui, HI
Las Vegas, NV
San Francisco, CA
Hawaii, HI
Los Angeles, CA
Portland, OR
Sacramento, CA
San Diego, CA
Kauai, HI
Orange County, CA
San Jose, CA
Seattle, WA
Denver, CO
Washington, DC
Phoenix, AZ
New York, NY
Anchorage, AK
Chicago, IL
Miami, FL
Reno, NV
Top 20 Cities/Counties Observed
Total Phone
Numbers Linked
to Location
Figure 3. Frequency Distribution of Advertised
Phone Numbers by City/County
Circuits were observed at both the state and
city/county level to provide insight on inter-state and
intra-state movement trends. At the state level, bi-
coastal traffic to Hawai‘i was observed in the data with
data filtered at high-level thresholds (weighted edge of
18 or greater) to detect the most prominent circuits.
This included portions of the Western Circuit (WA,
OR, CA, NV, AZ, and CO) as well as links between
Hawai‘i and New York, DC, Florida, and Texas
(Figure 4). Links between the Western Circuit states
and East coast states persisted with the removal of
Hawai‘i (sample source) indicating a high volume of
movement between those states (Figure 5). These
findings suggest that traffickers are using both ground
and airways to extend the reach of their networks.
Prevention and deterrent activity has primarily focused
on the use of roadways with interventions being
established at truck stops [1].
Figure 4. The Western Circuit and Bi-Coastal
Trends including Hawai‘i
Figure 5. The Western Circuit and Bi-Coastal
Trends with Hawai‘i Removed
In order to detect regional movement patterns, data
was filtered by region and at reduced thresholds to
observe an eastern circuit. With data filtered at a
weighted edge of at least eight, hub states of Florida,
Pennsylvania, New York, New Jersey, and
Massachusetts were observed (Figure 6, left hand side).
When thresholds were reduced movement patterns
between the hub states become apparent, making
greater portions of the eastern circuit visible (Figure 6,
right hand side). The trends observed were consistent
with known trafficking hubs and circuits based on
retrospective law enforcement data. In contrast with
retrospective data, our method could provide a
prospective tool for law enforcement and service
providers to identify trafficking activity in advance and
deter this activity.
Figure 6. Eastern Circuit
A city-level analysis was conducted to achieve
finer granularity. An analysis of the micro-circuits
within the top 10 hub states identified above was
completed. California’s micro-circuit included
advertisements in 30 different locations. Filters were
applied to identify the most prominent circuit, which
included movement between San Francisco,
Sacramento, San Jose, Los Angeles, Orange County,
and San Diego (Figure 7). Movement trends within
Nevada are predominantly between Las Vegas and
Reno. Movement patterns within Oregon were spread
across eight advertised locations. When filters were
applied to reduce noise the persistent circuit included
movement trends between Portland, Salem, and
Eugene. The observed micro-circuit within Texas
encompassed 13 advertised locations with heaviest
movement trends between El Paso, Abilene, Waco,
Austin, San Antonio, Houston, and Corpus Christi. The
micro-circuit in Washington that carried the highest
volume of advertisements included movement between
Figure 7. California Micro-circuit
Seattle, Yakima, and the Tri-cities. These findings
provide insight to the movement trends of potential
trafficking activity at the local level.
5 Discussion
Trafficking networks are increasingly using
technology to facilitate their activity. Research into
such uses is increasing, but much work is needed. The
intent of this study was to observe the types of data
available in online adult service advertisements that
offer indicators of trafficking and to identify ways to
transform this data into meaningful information that
can be used to disrupt potential criminal activity. The
primary focus was to obtain data to identify movement
trends using network analysis methods. By integrating
data from online classifieds and data obtained from
phone number analyses, it is possible to detect
indicators of sex trafficking and map patterns of
movement of potential traffickers or victims. As stated
above, the presence of sex trafficking indicators does
not prove trafficking is present, but it does identify
high-risk ads requiring further investigation.
Understanding the ways this community is using the
Internet (hub sites, important data fields, and how to
use that data) provides researchers with insight on how
to automate this process. Also, analysis of online
classifieds and the identification of advertisements
with potential links to trafficking have policy
implications in terms of websites facilitating criminal
activity. Most of the ads analyzed contained explicit
content obviously advertising prostitution with very
few ads attempting to hide the nature of the content.
Aside from issues pertaining to human trafficking,
websites hosting these types of ads are aiding illegal
actions. The Communication Decency Act of 1996
relieves websites from liability of third-party content.
This raises concerns about accountability and shared
liability of content posted on websites.
Tracking phone numbers proved to be an effective
method for detecting movement, as the phone is the
means of connecting the purchaser with the product. It
is also an integral element of product branding tied to
the provider’s online reputation through customer
reviews. The consistency in findings across a series of
studies and the ability to observe known circuits using
this methodology illustrates its effectiveness. The
ability to observe covert network activity can be
exploited to identify vulnerabilities to disrupt the
network. Federal efforts to combat human trafficking
call for an assessment of trafficking trends within the
U.S. The goal is to provide law enforcement and
service providers with the information they need to
more effectively manage and deploy resources [27].
Currently, law enforcement data on human trafficking
provides retrospective information of criminal activity.
Data is based on the number of arrests. The above
method outlines a way to capture movement trends of
potential trafficked persons prior to criminal action,
allowing for a more proactive approach to law
enforcement intervention.
Such a method would require further work to
develop, but might function as follows. Website
“scraping” software would be directed at online
classified sites, and natural language processing tools
could be used to identify pages bearing potential sex
trafficking indicators. These pages might be presented
to a human analyst who makes “include/don’t include”
judgments on a series of candidate pages. Phone
numbers and locations would then be extracted from
the selected corpus of advertisements, and aggregated
into a provider-location network. This network would
be automatically folded into a location-location
network in a manner constrained by transportation
network routes (highways and flights, e.g., travel
between Kahului and Lihue generally requires flying
through Honolulu). Identification of highly weighted
routes could facilitate the allocation of law
enforcement resources in general, and the
advertisements associated with the end-points of highly
weighted routes could be retrieved from the original
data to identify time windows for anticipated
movement of providers with specific advertised
characteristics between specific locations.
The observance of traffickers using both roadways
and airways to expand their markets presents clues to
potential intervention points via bottlenecks in the
supply flow. Traffickers may be harder to detect when
traveling along roadways. However, the use of airways
is of higher risk for traffickers and providers due to the
security measures implemented at airports. Air travel
requires identification and high levels of security
screening. Training security personnel to screen for
indicators of potentially trafficked persons could pose
an effective intervention strategy. Similar strategies are
in place for the prevention of drug trafficking in order
to cut off the supply flow.
6 Limitations and Future Directions
Limitations to this study include the manual
collection and processing of data, which is quite time
intensive and leaves room for error as some ads are
deleted or reposted. The methods used suggest
potential trafficking activity, but further analysis is
needed to definitively state that human trafficking is
present. The study observed the movement of an
advertised phone number along a circuit, but there is
not enough data to identify whom the phone belongs to
(trafficker or providers). Also, some numbers were
advertised in multiple cities on the same day, which
would require further investigation to identify exact
location of trafficker or provider. However, the
information obtained is useful in identifying movement
trends.
Presently the networks are based on temporal
adjacency but not sequence (e.g., a phone number
appearing at location A and then location B will
generate the same link as B followed by A). In future
studies, dynamic network analysis [20] will be applied
to analyze trail data (distinguishing sequences of
locations) in order to address the following questions:
What type of temporal information can be gathered
using open Internet sources? Can trends in movement
be observed using temporal data (i.e. rotations, average
length of stay in location, etc)? Can trafficker and/or
provider networks be identified using open Internet
sources?
The present study is built on advertisements
posted in Hawai‘i, so only includes data on persons
who appear in Hawai‘i, although we detected and
analyzed their appearance in other states as well. Data
collection will be expanded to include multiple starting
points in future studies. Multiple starting points may
present different regional trends and provide insight to
a larger more complex network. The exploration of
alternate data sources examining online human
trafficking activity beyond Backpage is also needed.
Online recruitment activity is another area needing
further exploration. The use of the Internet for
recruitment is a growing trend with recruitment activity
being reported in mainstream sites such as Facebook
and Twitter [7]. Further examination of these sites to
gain an understanding of how they are being used and
how potential victims are being targeted would provide
clues on how to disrupt and/or prevent this activity.
7 Acknowledgements
The second author was supported by NSF Award
0943147. The views expressed herein do not
necessarily represent the views of NSF.
8 References
[1] Operation Broken Silence. “The Nashville Backpage
Report: An Analysis of the Online Commercial Sex
Industry and Human Trafficking in Tennessee”, 2012,
Operation Broken Silence, available online
http://www.operationbroken silence.org/wp-
content/uploads/2012/01/Nashville-Backpage -
Report.pdf
[2] Belser, P.,de Cock, M., and Mehran, F. “An ILO
Minimum Estimate of Forced Labour in the World”,
2005, International Labour Office, Geneva.
[3] Mizus, M., Moody, M., Privado, C., Douglas, C.A.,
Collins, J.M., Mata, E., Dixon, L., and Stachowski, R.,
"Germany, Us Receive Most Sex-Trafficked Women",
Off Our Backs, 33(7/8), 2003.
[4] Judge, A., & Murphy, J. “Commercial exploitation of
children and adolescents in the US: A call to action for
psychology and law”. American Psychology-Law
Society Newsletter, 31(3), 2011, pp.24-28.
[5] TVPA. “Trafficking Victims Protection Act of 2000”,
2000, U.S. Department of State, available online
www.state.gov/documents/organization/10492.pdf
[6] boyd, d., Casteel, H., Thakor, M., & Johnson, R. “Human
Trafficking and Technology: A Framework for
Understanding the Role of Technology in the
Commercial Exploitation of Children in the U.S.”, 2011,
Microsoft Research, available online
http://research.microsoft.com/enus/collaboration/focus/e
ducation/htframework2011.pdf
[7] Latonero, M., "Human Trafficking Online: The Role of
Social Networking Sites and Online Classifieds",
available at SSRN 2045851, 2011.
[8] Hughes, D. “Use of the Internet for Global Sexual
Exploitation of Women and Children”, 1998, University
of Rhode Island, Women’s Studies Program, The
Coalition Against Trafficking in Women, available
online http://www.uri.edu/artsci/wms/hughes/pubtrf
rep.htm
[9] Hughes, D. M. “The Internet and Sex Industries: Partners
in Global Sexual Exploitation”, Technology and Society
Magazine, IEEE, 19(1), 2000, pp. 35-42.
[10]Latonero, M. “Technology and Human Trafficking: The
Rise of Mobile and the Diffusion of Technology-
Facilitated Trafficking”, available at SSRN 2177556,
2012.
[11] Sparrow, M.K., "The Application of Network Analysis
to Criminal Intelligence: An Assessment of the
Prospects", Social networks, 13(3), 1991, pp. 251-274.
[12] Klerks, P., "The Network Paradigm Applied to Criminal
Organizations: Theoretical Nitpicking or a Relevant
Doctrine for Investigators? Recent Developments in the
Netherlands", Connections, 24(3), 2001, pp. 53-65.
[13] Stoecker, S. “Demand. A comparative examination of
sex tourism and trafficking in Jamaica, Japan, the
Netherlands, and the United States”, 2009, Shared Hope
International, available online http://sharedhope.org/
resources/
[14] Estes, R.J., and Weiner, N.A., The Commercial Sexual
Exploitation of Children in the Us, Canada and Mexico:
Full Report (of the Us National Study), University of
Pennsylvania, School of Social Work, Center for the
Study of Youth Policy, 2001.
[15]Hanna, C., "Somebody's Daughter: The Domestic
Trafficking of Girls for the Commercial Sex Industry
and the Power of Love", William and Mary Journal of
Women and the Law, 9(1), 2002.
[16] Kreyling, S., West, C., and Olson, J., "Technology and
Research Requirements for Combating Human
Trafficking", Pacific Northwest Laboratory Richland,
Washington, Tech. Rep. PNNL-20258, 2011.
[17] Smith, L., Vardaman, S., & Snow, M. “The National
Report on Domestic Minor Sex Trafficking: America's
Prostituted Children”, 2009, Shared Hope International,
available online http://sharedhope.org/ resources/
[18] TIP. “Trafficking in Person's Report 2012”, 2012, U.S.
Department of State, available online
http://www.state.gov/j/tip/rls/tiprpt/2012/index.htm
[19]Lee, H., and Lee, J., "Computer-Mediated
Communication Network: Exploring the Linkage
between Online Community and Social Capital", New
Media & Society, 12, 2010, pp. 711–727.
[20] Carley, K.M., "Dynamic Network Analysis", in (Editor,
R. Breiger, K.M. Carley, & P. Pattison): Book Dynamic
Network Analysis, Comittee on Human Factors,
National Research Council, 2003, pp. 133-145.
[21] Whitaker, M. “Prostitution ads: Record visits, $3.2
million sold”, 2012, AIM Group, available online
http://aimgroup.com/2012/04/23/prostitution-ads-
record-visits-3-2-million-sold/
[22] Mcclain, N.M., and Garrity, S.E., "Sex Trafficking and
the Exploitation of Adolescents", Journal of Obstetric,
Gynecologic, & Neonatal Nursing, 40(2), 2011, pp. 243-
252.
[23] UNODC. “United Nations Office on Drugs and Crime,
Human trafficking indicators”, 2012, UNODC, available
online http://www.unodc.org/pdf/HT_ indicators_ E_
LOWRES.pdf
[24] Polaris. “Domestic minor sex trafficking: The criminal
operations of the American pimp”, 2012, Polaris
Project, available online
http://www.polarisproject.org/resources/ resources-by-
topic/sex-trafficking
[25] Operation Broken Silence. “The Cleveland Backpage
Report: An Analysis of Human Trafficking and the
Online Commercial Sex Industry in Northeast Ohio”,
2011, Operation Broken Silence, available online
http://www.operationbrokensilence.org/wp
content/uploads/ 2011/10/Cleveland-Report-Final-
Release1.pdf
[26]ORA. “Organizational Risk Analyzer”, n.d.,
Computational Analysis of Social and Organizational
Systems (CASOS), available online
http://www.casos.cs.cmu.edu/projects/ora/
[27] WHOPS. “Fact sheet: The Obama Administration
Announces Efforts to Combat Human Trafficking at
Home and Abroad”, 2012, The White House Office of
the Press Secretary, available online
http://www.whitehouse.gov/thepressoffice/2012/09/25/f
act-sheetobamaadministration
