Big Data For Criminal Investigation

Chitra Lakhani, Vikas Janu

JK Lakshmipat University Jaipur 302026 India

ARTICLE INFO ABSTRACT

Keywords: In this paper, we will be giving a brief idea about the work that is done related to big data in the field
Big data of legal and law enforcement. Starting with a brief introduction to big data and big data in legal and
Criminal Investigation law enforcement. After that, background information about big data and survey on big data in the field
Legal of legal and law enforcement. We discuss some of the progress that has happened in the field until
Data Mining now, as per the data mining techniques and also what are the problems faced regarding the same.

1. Introduction to make informed decisions. It involves the use of advanced

tools and techniques such as data mining, machine learning,
On a daily basis, people from all over the world post a and predictive analytics to extract insights from big data.
huge amount of information on various social media plat-
One of the main challenges of big data is its manage-
forms. Police have learnt to utilize social media to obtain
ment, processing, and analysis. Traditional data processing
information on suspects, witnesses, or events in hopes of tools and techniques are often inadequate for handling big
resolving crimes and stopping planned criminal behavior. A data, which requires specialized software and hardware in-
certain amount of this intelligence is of value to authorities in frastructure, as well as skilled data scientists and analysts.
law enforcement. Although police can use search warrants to Another challenge of big data is ensuring its quality and
gain access to non-public social media data, they frequently
accuracy, as well as protecting its privacy and security. As
find the material they need by searching public forums,
big data becomes increasingly important in various indus-
where access is not restricted by the need for a search tries, there is a growing need for standards, regulations, and
warrant. best practices to ensure the ethical and responsible use of
The nature and scope of police usage of social media are data.
evolving as we go into the big data era. Police departments Despite the challenges, big data has the potential to
have decided to experiment with analytics on social media
revolutionize various fields such as healthcare, finance, mar-
for the sake of surveillance, profiling, and predictive ana-
keting, and government. It can help organizations to make
lytics, helped by start-ups providing data analytics targeted better decisions, improve their operations and services, and
to law enforcement demands. These actions bring up fresh create new opportunities for innovation and growth.
concerns about social fairness and privacy. They also bring
up crucial questions of accountability and transparency. By 2.2. Data Mining
doing thus, they draw attention to the expanding connections Data mining is a rapidly expanding field that comes at
among private and other legal "disciplines." The reports in the intersection of many disciplines like statistics, database,
the fall of 2016 that exposed the widespread utilization of machine learning, etc. It combines statistical modelling, data
social media analytics by American police forces serve as storing and approaches in AI. The main goal is to learn or
the backdrop for this paper’s examination of these concerns. predict the future, from the current structured or unstruc-
It identifies the social justice and privacy concerns created tured data, analysing the trends in the present data, and
by these actions and looks at normative and transparent learning from it. There are models introduced by different
frameworks for using such technologies. statisticians that can be used to examine these trends or
relationships between the variables in a database in order to
2. Background create prediction models and other useful models.
AI being one of the most important of those sub-fields to
2.1. Big Data set the base for data mining, has many algorithms developed
Big data refers to extremely large sets of data that can to automate learning from data. These algorithms can lay
be analyzed for patterns, trends, and associations using ad- a solid foundation for the development of various prediction
vanced computing technologies. The term "big data" has models such as criminal activity detection, criminal behavior
emerged as a result of the massive growth in the volume, detection, etc. in the field of criminal activity. There are
velocity, and variety of data generated by various sources indeed steps being taken for the same to try and stop criminal
such as social media, mobile devices, sensors, and other activities from occurring.
digital technologies.
Big data analytics is the process of examining large 2.3. Legal and Law Enforcement
and complex data sets to uncover hidden patterns, unknown The application of big data in law enforcement has seen
correlations, and other useful information that can be used a significant increase in recent years as the vast amount
ORCID (s): of digital data available continues to grow exponentially.

Alok Kumar et al.: Preprint submitted to Elsevier Page 1 of 5

 CS1223: Google Cloud Report

Law enforcement agencies can access various types of data,

including surveillance footage and social media activity, to
aid in their decision-making processes. This development
has led to exciting possibilities such as the use of predictive
analytics to identify potential criminal activity and advanced
algorithms for facial recognition and biometric identifica-
tion. However, the use of big data in law enforcement is not
free from challenges and controversies. Utilizing advanced
data analytics techniques can enhance the effectiveness and
efficiency of investigations, allowing law enforcement agen-
cies to quickly process large amounts of information to A prototype system was introduced by Jennifer Schroeder
detect patterns and anomalies that may have gone unnoticed and her group called CrimeLink Explorer for enabling
through traditional investigative methods. Yet, one of the automatic crime link analysis. The approaches suggested in
main concerns with using big data is the potential for bi- their system by them are co-occurrence analysis, a heuristic
ased algorithms that may perpetuate systemic injustices if technique for discovering links between crime items, and
improperly designed or trained on biased data. Moreover, shortest path algorithm for reaching the goal in minimum
the collection and use of personal data raise concerns about time. These approaches aim to integrate the investigative
privacy and civil liberties, as well as the possibility of data domain knowledge in the system for automatically deter-
misuse by law enforcement agencies. Therefore, it is crucial mining the strength of association. For this, the system will
that the ethical and legal implications of the use of big data use event similarity, time-based relation and connection, and
in law enforcement are examined. Policymakers must weigh other variables to identify relations between terrorist attacks
the potential risks and benefits of using big data in law by extracting a criminal’s behavior and any other pattern.
enforcement and establish safeguards to ensure responsible Another main element that helps criminals to easily
and ethical usage. This paper will discuss the main ethical widen their horizon of impact is through social media.
and legal considerations linked to the use of big data in law Regarding this, another tool is introduced, social network
enforcement and propose recommendations to responsibly analysis (SNA). This is also a tool with a link analysis
and effectively use big data in the criminal justice system. approach implemented in it. It uses the level with which the
suspects on the social network are related to each other in
proximity, thinking, etc.
3. Data Mining Technologies
On the basis of current research and practice, it is found
As the amount of crime occurring nowadays increases, that we can use link analysis to explore associations among a
there is a requirement for better technology and innovative large number of nodes and with a large number of features to
ideas to fight off these crimes. This section explains some of link them. Link analysis is usually useful for the analysis of
the technologies in data mining that are or can be used in the structured data but can also be used to examine unstructured
field of crime. data by integrating some text mining methods.
3.1. Link Analysis 3.2. Text Mining
Link analysis is a data mining technique that shows the Also known as text data mining, is the process of ex-
structure of the data in the form of a collection of connected tracting important and hidden knowledge from text doc-
or linked objects. Linking the data in an interconnected form uments. Text mining focuses on the concept or theme of
is only the beginning. These links between the data can be the data given instead of specific words, classifying them
used to identify relationships between objects that would accordingly, and using these classifications to help ana-
otherwise look unlinked from a simple point of view. This lysts and investigators in finding useful data or information
technique can be used to find links in relation to terrorism, within a large amount of documents. Text mining tools and
fraud, drug dealings and other matters. This is the first procedures provide appropriate links among a variety of
method in the list of techniques in data mining that can be crucial textual information for learning new information and
used in exploring, assembling, detecting and analysing huge choosing the best course of action. Authors in [1] suggested
networks of pupil data through their jobs, vehicles, bank using text mining to uncover criminal networks from the
accounts, addresses, connections and other features. Linked gathered textual materials. The suggested method primarily
data is visualised in the form of a graph with several nodes finds relevant information for criminal research before visu-
being people of interest and the links between them defining alising the retrieved criminal network for analysis. Similar
their relationship with each other. research is shown in [2] where information retrieval is done
Over the past few years, various research has used link from news reports regarding crimes. Another work in [3],
analysis to examine huge social networks. The volume of to police investigations, a comparison of the usability of
data does not matter when using link analysis to find rela- emergent self-organizing maps (ESOM) and multidimen-
tions between all the suspects. What matters is their relation sional scaling (MDS) as text exploration tools was done for
to each other, how they are connected, what they are doing finding hidden information from unstructured data of police
to each other.

Alok Kumar et al.: Preprint submitted to Elsevier Page 2 of 5

 CS1223: Google Cloud Report

records. Likewise, another technique was presented in [4], network analytics because they can effectively integrate the
primarily using NLP technique to find entities like occupa- elusive qualities of human reasoning with the perfect mem-
tion, people, addresses etc, by matching words or sentences ory of computers, compulsive thoroughness, precise logic
from the already available text data. Then, those words and commutation.
or sentences are compared within the criminal network to Using ANN, a model was proposed in [12], where the
find any associations or trends. It is feasible to recognise previous knowledge of the behavior of a criminal can be
dishonest people using text mining techniques based on used to predict the future crime location. The authors used
linguistic characteristics. For instance, text mining and data criminal classification to retrieve the most accurate pattern.
mining techniques were used in [5] to find bogus items in a A number of ANN tools have been developed over the years,
textual content. A rule-based approach was created by Yin to having user friendly interfaces and having the ability to
capture sentiment features, where contextual elements were automatically change, according to their environment, their
found by comparing textual postings to a window of nearby internal structure.
textual posts in [6]. Additionally, SVM was used in the
text mining technique to categorise abusive posts on online
social media. Furthermore, perilog is a tool introduced by
NASA that can extract relevant data by context from any
text document or phrases in [7]. It was originally created
as part of FAA’s Aviation Safety Reporting System, whose
implementation was done to look for the primary reason for
the crash of an airline in [8].

3.3. Artificial Neural Network

An artificial neural network imitates the working of the
human brain’s neurons, as a model of several hundreds and
thousands of computational units. Some of these inbuilt,
through software, abilities of ANN are memory, learning and
decision making. Forecasting possible criminal patterns on 4. Criminal Investigation Implementation
the basis of observations on ongoing criminal activity using 4.1. Introduction
big data analytics, using the current sample to get trends and In recent years, the importance of big data in the field of
predict new trends. A NN can be used to build a tool which law enforcement has grown significantly. Big data refers to
can help in behavioural and psychological examination of vast amounts of information generated from various sources,
a suspect. An example is the project in [9] which is a psy- including social media, surveillance cameras, and other tech-
chological profiling system built on ANN. ANN models are nologies. Law enforcement agencies are utilizing big data
more capable than standard database-oriented approaches in to enhance their ability to prevent and investigate crimes.
link analysis. Investigators can use it to retrieve important For example, they can utilize data analytics tools to iden-
information from the police narrative report textual data. The tify patterns in crime data, such as the time, location, and
Coplink project in [10] is the extraction of name entities nature of crimes, enabling them to allocate resources more
based on ANN. Another recent study in [11] based on efficiently. Furthermore, they can monitor social media and
ANN is the retrieval of associations based shortest path other online sources for potential threats or criminal activity.
algorithms. Nevertheless, the application of big data in law enforcement
The shortest path technique was used to find the strength raises concerns regarding privacy and civil liberties. Critics
of an association in the criminal network using ANN. De- argue that collecting and analyzing enormous amounts of
spite the limitations imposed by the complex properties of data can lead to profiling and discrimination, and that using
criminal data on conventional security and criminal investi- automated tools to analyze data may lead to false positives
gation techniques, ANNs can significantly advance criminal and other errors. To address these concerns, law enforcement

Alok Kumar et al.: Preprint submitted to Elsevier Page 3 of 5

 CS1223: Google Cloud Report

agencies are establishing policies and procedures to ensure

that the use of big data is transparent, ethical, and account-
able. This includes implementing measures to protect the
privacy and civil liberties of individuals and guaranteeing
the accuracy and reliability of data analytics tools. To sum
up, while the use of big data has the potential to improve
the efficiency of law enforcement agencies, it should be
balanced against concerns about privacy and civil liberties.
Law enforcement agencies must consider the ethical and
legal implications of utilizing big data and ensure that its
usage is responsible and accountable. and the appropriateness of using results in decision-making
are some of the issues. Even though all these issues affect all
4.2. Tools and Resources empirical methodologies, big data analytics has the potential
Agencies can gather information from many resources to be more effective, opaque, and mythical than traditional
for examining criminal networks, which can include service approaches. The main objective of this article is to demystify
logs, location based networks, call records, financial state- some tactics so that we may understand how and when to
ments and previous crime reports. Primary means for law use them properly. While we cannot ban people from using
enforcement agencies are call records, bank statements and statistics, we can actively discuss the kind of interpretations
call records. Recently, messages and emails, under text data, that can and should be made as well as the impact that these
have also been used for examining to disclose close relations inferences should have on various types of decisions.
and links between groups. Several social media applications The methods used by attorneys and law enforcement to
have also been used for positioning analysis, which has draw conclusions and reach judgements are not unbiased,
the motive of finding closeness between two participants which is also the case with many technologies. It is notable
rather than between their geographical location. Some of the that there has been a shift away from using court decisions as
tools used for big data are HDFS, pig, hive, MapReduce, precedents to build doctrinal arguments and towards seeing
Cassandra, Avro, etc. them as data that can be analyzed using statistical methods.
The transition to "predictive policing," in which deployment
choices are informed by statistical forecasts, is also accept-
able. Some people may find a trend in this direction to be
appealing due to the strength of big data analysis and its
prospective efficacy in discovering correlations. But, one
would need to go through the police and lawyers’ historic
resistance to changing their ingrained ways of thinking.

4.4. Progress
The integration of big data in law enforcement has been a
rapidly advancing field, increasingly utilized by law enforce-
ment agencies globally. The following are some examples of
how big data has been implemented in law enforcement, as
well as the advancements made:
4.3. Method - Crime Prevention: Law enforcement agencies are lever-
The above mentioned data mining techniques are few of aging big data to prevent crimes before they happen. They
the ones that can be used to get information about crimes use data analytics to study previous crimes, predict where
and criminals from the available data. Cluestering, block- and when future crimes are likely to occur, and then allocate
mining, association rules are some of the techniques that resources more efficiently to proactively prevent crimes.
can be used for this purpose. Some authors have proposed - Investigation: Big data is also used for investigations.
solution that implement most of these techniques and can By analyzing data from social media, surveillance cameras,
be practically implemented after overcoming their respective and other sources, law enforcement agencies can collect ev-
fields’ challenges. We can build systems with these technolo- idence and identify suspects more efficiently and accurately.
gies, feed data to it, and then the system will train to predict - Predictive Policing: Predictive policing is a method that
future trends. uses data analysis to identify areas with a high likelihood
of crime. Law enforcement agencies use big data to build
Conclusion predictive models that determine where and when crimes are
Big data applications in law run the risk of being char- most likely to occur, enabling them to deploy resources more
acterized as either inevitable or impractical. Understanding effectively.
the techniques used, the biases introduced when selecting a - Sentencing: Big data is also used to inform sentencing
method for machine learning, the type of inferences formed, decisions. By studying data on past sentencing decisions,

Alok Kumar et al.: Preprint submitted to Elsevier Page 4 of 5

 CS1223: Google Cloud Report

law enforcement agencies can identify disparities and biases scaling: A comparative study on domestic violence, Applied Soft
and develop more equitable sentencing guidelines. Computing 11 (2011) 3870–3876.
- Real-Time Analytics: Real-time analytics involves an- [4] R. Lee, Automatic information extraction from documents: A tool for
intelligence and law enforcement analysts, in: Proceedings of 1998
alyzing data as it is produced to provide law enforcement AAAI Fall Symposium on Artificial Intelligence and Link Analysis,
agencies with current information on possible threats and volume 23, AAAI Press Menlo Park, CA, 1998.
criminal activity. This allows them to respond faster and [5] C. M. Fuller, D. P. Biros, D. Delen, An investigation of data and text
more effectively to emerging situations. mining methods for real world deception detection, Expert systems
In conclusion, the application of big data in law en- with applications 38 (2011) 8392–8398.
[6] D. Yin, Z. Xue, L. Hong, B. D. Davison, A. Kontostathis, L. Edwards,
forcement has made significant strides, and its usefulness Detection of harassment on web 2.0, Proceedings of the Content
will only continue to grow with advancements in technology Analysis in the WEB 2 (2009) 1–7.
and data analytics. However, it is critical that ethical and [7] M. W. McGreevy, Using Perilog to Explore" Decision Making at
legal considerations are made to ensure responsible and NASA", Technical Report, 2005.
accountable use of big data in law enforcement. [8] C. Billings, J. Lauber, H. Funkhouser, E. Lyman, E. Huff, NASA
aviation safety reporting system, Technical Report, 1976.
[9] M. Strano, A neural network applied to criminal psychological
4.5. Problems Faced profiling: An italian initiative, International Journal of Offender
Working with big data in law enforcement comes with Therapy and Comparative Criminology 48 (2004) 495–503.
several challenges that need to be addressed to ensure re- [10] T. M. Li, M. Chau, P. W. Wong, P. S. Yip, A hybrid system for
sponsible and effective use of the technology. The following online detection of emotional distress, in: Intelligence and Security
Informatics: Pacific Asia Workshop, PAISI 2012, Kuala Lumpur,
are some of the problems that law enforcement agencies face Malaysia, May 29, 2012. Proceedings, Springer, 2012, pp. 73–80.
when working with big data: [11] J. J. Xu, H. Chen, Fighting organized crimes: using shortest-path
- Privacy concerns: The collection and use of large algorithms to identify associations in criminal networks, Decision
amounts of personal data, such as information from social Support Systems 38 (2004) 473–487.
media and surveillance cameras, raise privacy concerns. [12] K. Dahbur, T. Muscarello, Classification system for serial criminal
patterns, Artificial Intelligence and Law 11 (2003) 251–269.
Law enforcement agencies must balance the need for public
safety with individual privacy rights.
- Accuracy and reliability of data: The accuracy and
reliability of big data are crucial to its effectiveness in A. IPR Certificate
law enforcement. Errors in data collection, processing, or We, [Your Name] and [Your Partner Name], with this
analysis could lead to false positives, which can have severe certify that the project work submitted by us entitled [Project
consequences. Title] to our supervisors, Dr. Alok Kumar and Dr. Utsav
- Data security: The vast amount of data collected by Upadhyay, in partial fulfillment of the requirements for the
law enforcement agencies needs to be kept secure to protect course is a bonafide work carried out by us and has not been
against data breaches or unauthorized access. previously submitted to any other course. We further certify
- Bias and discrimination: There is a risk that big data that no part of this work shall be published, reproduced,
analysis could lead to profiling or discrimination against or distributed in any form without the prior permission of
certain groups based on factors such as race or gender. our supervisors. We understand that any such unauthorized
- Technological limitations: Law enforcement agencies use of the project work may be considered a violation of
may face technological limitations in terms of infrastructure, academic ethics and result in severe penalties. We also affirm
data processing, and data storage capacity. that the project work has been carried out under the ethical
- Legal and ethical considerations: The use of big data standards and guidelines set forth by our supervisors. We
in law enforcement also raises legal and ethical considera- acknowledge that our supervisor has the right to make any
tions, such as data privacy, transparency, and accountability. modifications or revisions to the project work that may be
Overall, the use of big data in law enforcement presents deemed necessary. We also agree to abide by any additional
significant challenges that must be addressed to ensure that terms and conditions as stipulated by our supervisor.
its use is responsible, ethical, and effective. Law enforcement
agencies need to implement policies and procedures that Date: 20th May, 2023
address these challenges and ensure that the use of big data
is transparent and accountable. Signature of Student:
Chitra Lakhani
References 2020BTECHCSE024
JK Lakshmipat University, Jaipur
[1] R. Al-Zaidy, B. C. Fung, A. M. Youssef, Towards discovering
criminal communities from textual data, in: Proceedings of the 2011
ACM Symposium on Applied Computing, 2011, pp. 172–177. Signature of Student:
[2] Y.-H. Tseng, Z.-P. Ho, K.-S. Yang, C.-C. Chen, Mining term networks Vikas Janu
from text collections for crime investigation, Expert Systems with 2020BTECHCSE083
Applications 39 (2012) 10082–10090.
JK Lakshmipat University, Jaipur
[3] J. Poelmans, M. M. Van Hulle, S. Viaene, P. Elzinga, G. Dedene, Text
mining with emergent self organizing maps and multi-dimensional

Alok Kumar et al.: Preprint submitted to Elsevier Page 5 of 5