This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS 1
Finding Emergent Patterns of Behaviors in Dynamic
Heterogeneous Social Networks
Benjamin W. K. Hung , Anura P. Jayasumana, Senior Member, IEEE , and Vidarshana W. Bandara
Abstract—The search in graph databases for individuals or
entities undertaking latent or emergent behaviors has applica-
bility in the areas of homeland security, consumer analytics,
behavioral health, and cybersecurity. In this setting, even partial
matches to hypothesized indicators are worthy of further investi-
gation, and analysts in these domains aim to identify and main-
tain awareness of entities that either fully or partially match the
queried attributes over time. We provide a comprehensive version
of a graph pattern matching technique called Investigative Search
for Graph Trajectories (INSiGHT) to find emergent patterns of
behaviors in networks and tailor the application to detecting rad-
icalization in the homeland security domain. To enable analysts’
accounting of recurring behavioral indicators and the recency of
behaviors as the imminence of a threat, we provide parameterized
methods to score multiple occurrences of indicators and to
dampen the significance of indicators over time, respectively.
Additionally, we provide an indicator categorization scheme and
a match filtering technique to ensure that partial matches to
the most salient indicators are identified while reducing the
number of false positives. Furthermore, since individuals may be
radicalized in small groups or be involved in collective terrorist
plots, we introduce a non-combinatorial neighborhood matching
technique that enables analysts to use INSiGHT to identify
potential query matches from clusters of individuals who may
be operating in conspiracies. We demonstrate the performance
of our approach using a synthetic radicalization data set and a
large, real-world data set of the BlogCatalog social network.
Index Terms— Graph pattern matching, graph streams, inves-
tigative graph search, pattern matching trajectories.
I. I NTRODUCTION
S
EARCH in graph databases for individuals with specific
types of connections or attributes is an increasingly rich
research area. In particular, pattern matching in graphs has
been studied for use in social search and recommender systems
in [1], [6], [11], [13], [26], [30], [31], [35], and [40]. However,
there are several shortcomings in current approaches when
applied to the search for individuals or entities undertaking
latent behaviors, which are hidden or emergent activities
exhibited by an entity [14]. This kind of search problem is
Manuscript received January 15, 2019; revised June 9, 2019 and July 29,
2019; accepted August 11, 2019. This work was supported in part by
the U.S. Department of Justice, Office of Justice Programs/National Insti-
tute of Justice, under Award 2017-ZA-CX-0002. (Corresponding author:
Benjamin W. K. Hung.)
B. W. K. Hung and A. P. Jayasumana are with the Depart-
ment of Electrical and Computer Engineering, Colorado State Univer-
sity, Fort Collins, CO 80525 USA (e-mail: benjamin.hung@colostate.edu;
anura.jayasumana@colostate.edu).
V. W. Bandara is with JDA Software Group, Inc., Irving, TX 75062 USA
(e-mail: vida.bandara@jda.com).
Digital Object Identifier 10.1109/TCSS.2019.2938787
particularly appropriate for law enforcement and intelligence
analysis. In this setting, even partial matches to hypothe-
sized indicators are worthy of further investigation. Moreover,
the aim is to identify and maintain awareness of entities that
either fully or partially match the queried attributes over time.
As the underlying entity-level data are dynamically updated
with behaviors and attributes, the goal is to find all those with
the emergent pattern of behaviors.
Emergent pattern detection is a process by which analysts
discover the occurrence and logically connect an entity’s set of
activities over time. Analysts identify a hypothesized pattern
of behavior and subsequently employ technological means
to detect characteristic behaviors which match the pattern in
order to provide an early warning. Emergent pattern detection
is important in domains such as homeland security, consumer
analytics, behavioral health, and cybersecurity because of the
compelling interest to detecting the presence of an individ-
ual’s latent behavior utilizing time-stamped indicator data.
To prevent future terrorist attacks, law enforcement agencies
recurringly assess the individual risk of a large number of
individuals for the likelihood of violence as they progress
along a dynamic and phase-based radicalization process and
exhibit indicative behaviors or psychological states [3], [25],
[27], [28]. Similarly, in consumer analytics, businesses are
interested in an individual’s online activities and purchases
over time to track his or her place on the customer journey
and determine the potential for future purchases [9], [10], [42].
In behavioral health, family members and caregivers are inter-
ested in identifying those who may be exhibiting indicators of
suicide risk over time [24], [34], [36]. Lastly, in cybersecurity,
organizations continually seek to prevent insider threats by
detecting risk potential using performance-related and techni-
cal indicators recorded over time [4], [8].
In summary, emergent pattern detection utilizing longitudi-
nal characteristics and activities’ data is applicable in numer-
ous contexts. However, this challenge is not yet adequately
addressed by extant graph pattern matching approaches. There-
fore, we formulate a solution approach called investigative
graph search [18] that enables the search for and prioritization
of entities of interest that over time exhibit part or all of a
pattern of attributes or connections.
A. Basic Investigative Graph Search Problem
We start with a graph model consisting of nodes and edges.
The node types, constrained by a discrete number of node
2329-924X © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.