Crime Investigation System

Laboni Halder1, Sapna Jaiswal2, Shriya Tiwari3, Linda John4

1,2,3,4
Department of Information Technology, St. John College of Engineering and Management

Abstract— The number of criminal offense that are reported per day in India is increasing drastically.
The criminals commit crimes in really tactful ways. This makes investigation a more complex process.
Thus the police officers have to go through a lot of manual tasks to get a thread for investigation.
Security has become a primary concern. It is given high priority by the government worldwide with
the aim to reduce crime incidents. So we are proposing a system that will simplify the investigation
process and help in getting faster results. The system utilizes data mining methods for analyzing
criminal information to perform an effective investigation of the criminal acts thereby, automating the
process for police officers. Data mining plays an important role in prediction and analysis of criminal
information. It is the appropriate field to apply on high volume crime dataset and knowledge gained
from data mining approaches will be useful and support police force. A criminal dataset is created
which includes several different types of crime that are based on real data. Clustering is the process of
combining data objects into groups. The data objects of each group are similar to each other. But they
are very dissimilar when compared to objects of other groups. These groups will help in deriving
relations between the data objects. In this paper we use python to apply K-means clustering to perform
criminal analysis on crime dataset. K-means method is a widely used clustering technique that aims to
minimize the average squared distance between points in the same cluster. It offers great simplicity
and speed. The purpose of this study is to predict the offender by taking into consideration the known
characteristics of past criminal cases.
Keywords— FIR, Cluster, Crime Analysis and K- means.
I. INTRODUCTION
A report compiled by the World Health Organization stated that more than 50 percent of citizens
around the world have been victims of criminal activity. The National Crime Records Bureau of India
reported that a woman is assaulted in the country every three minutes. The police are not able to help,
as information about the crime does not reach them in time. With the number of criminal activity
increasing at such an appalling rate, it is evident that a method is required from the technical
community to control the current situation. This paper presents Crime Investigation System (CIS), a
website that can serve the need of both citizens as well as police officers. CIS deals with analyzing the
crime information and automate investigation for police officers by using data mining technology. To
sort large data sets data mining is the appropriate tool. It helps in identifying patterns and establish
relationships to solve problems through data analysis. The main purpose of the proposed system is to
ease the task of lodging complains thereby, motivating the victims to stand-up to criminals. The system
is more efficient as it ensures confidentiality and speeds up the investigation process. This paper
describes the development and usage of the application, as well as its technical implementation and
future prospects and proposals.

II. LITERATURE SURVEY

The literature review describes the current knowledge and the future possible findings for a particular
topic. It also describes the related theory and methodologies that can be included. It act as basis for
research and study on a particular domain. The literature review for this system includes the study and
description of 5 technical papers.

DOI:10.23883/IJRTER.2018.4091.CQLB6 32
 International Journal of Recent Trends in Engineering & Research (IJRTER)
Volume 04, Issue 02; February- 2018 [ISSN: 2455-1457]

1. E-Police System-FIR Registration and Tracking through Android Application

In [1] Archana Iyer, Prachi Kathale, Sagar Gathoo, Nikhil Surpam presented an approach to lodge the
FIR through android application via phone. The system needs an easily accessible android mobile.
Earlier the entire manual process was time consuming as the complainant had to physically go to the
police station numerous times. Thus the system provides easier way to lodge the FIR from anywhere.
The same also consumes a lot of money and energy. The technology used by this system is Android
which is an operating system for touchscreen mobile phones. The Android’s user interface allows
direct manipulation using touch, gesture, manipulate objects on the screen, virtual keyboard for textual
input. Also Eclipse which is an integrated development environment (IDE) is used for developing Java
applications. WAMP server is Windows web development environment, it allows one to create web
applications using Apache2, PHP for creating the dynamic web pages and MySQL for creating the
database.

2. Criminal Investigation and Identification System

In [2] Saurabh Zadikar, Jaan Momin, Datta Mule, Avinash Chavan presented an approach to lodge the
FIR online through mobile phone. The main purpose of this system is to provide easy way to file FIR
by victims and based on criminal data, storing criminal records which will help in further
investigations. Initially the victim registers FIR, gets OTP and login to do further process of
investigation. At police side the investigation process is done and the updates and alerts regarding to
case are sent to user. Also chat system is provided from police to police. During this investigation
process the criminal records are being updated. The database is used for storing the information of
criminal and thumb image for recognition. This process is done online, input image of thumb can be
taken by using biometric device. The system maintains details of criminal pattern once it is saved in
the database.

3. Cluster Based Zoning of Crime Info

In [3] Lalitha Saroja Thota, Mohrah Alalyan, AL-Otaibi Awatif Khalid, Fabiha Fathima, suggested
an approach of cluster technique of K-means algorithm on criminal dataset of India for crime analysis.
The crime dataset is developed by applying complex query on the Crime Info India database. The
crime profiling and zoning can be modelled with utilization of data mining. To construct cluster zones
which is based on K-means clustering technique the crime dataset is given as input to WEKA software.
The cluster technique builds a model of states with high, medium and low crime zones. The cluster
output of WEKA is passed manually as input to My Custom map, an online interactive map tool of
maps of India creates custom India map with the cluster zones of states. The custom maps displays an
overall crime profiles of states which helps police and law enforcement department to take additional
preventive measures to combat against crime and plan advanced investigation strategies. The crime
trend and zoning knowledge can also be helpful in cautioning police in levels of preventive actions.

4. An Enhanced Algorithm to Predict a Future Crime us Data Mining

In [4] Malathi. A, Dr. S. Santhosh Baboo presented an approach to predict the patterns of crime and
process to solve crimes. This proposed system uses MV Algorithm and Apriori Algorithm for filling
missing values. Also it identifies the crime patterns. Clustering algorithm is used for data mining to
help the crime prediction. For discovering the knowledge from crime records supervised learning
technique is used, this increases predictive accuracy. This entire process can help law enforcement
department to efficiently handle crime investigation. Knowledge discovery tool that is used to review
extremely large datasets and incorporate a vast array of methods for accurate handling of security
issues.

5. Efficient Algorithms to find Frequent Itemset Using Data Mining

In [5] Sagar Bhise,Prof. Sweta Kale presented an approach of Frequent Pattern growth algorithm to

@IJRTER-2018, All Rights Reserved 33

 International Journal of Recent Trends in Engineering & Research (IJRTER)
Volume 04, Issue 02; February- 2018 [ISSN: 2455-1457]

accomplish effectiveness and efficiency using Frequent Itemset Mining algorithm. This algorithm
classifies into two phases namely pre-processing phase and mining phase. The pre-processing phase
have to improve effectiveness and novel smart splitting technique for transformation of the database.
The data which is lost during splitting of transaction is taken as offset by the mining phase and then a
run time estimation method is calculated to find the actual support of item set in a given database.

III. PROPOSED SYSTEM

We have divided the entire system implementation into two stages. The first stage involves developing
a graphical user interface which will enable the users to interact with the system. Users will be able to
register FIR via the forms developed in this stage. In the second stage, a dataset of criminals is
prepared. This dataset contains information related to the criminals such as their physical features,
date-time, location and type of crime they committed. This stage also involves application of K-means
algorithm on the information provided by the user and based on the dataset that has been prepared a
prediction is done and a list of criminal/ criminals is generated. After that the victims are being notified
about the status of their report.

Figure 1 System Architecture

Phase 1
Users first register and then login on the website for lodging FIRs or complaints. These FIRs and
complaints can be viewed on the police side once it has been submitted by the user through the website.

Phase 2
The police has to register and then login in order to view FIRs and complaints. An automated
investigation will be then performed on those FIRs. Once an investigation is completed the police will
update the status of the FIR thereby, informing the complainant about it.

Phase 3
A criminal dataset is taken, it is filtered according to requirement, missing values are filled and then
analysis can be done. K-means clustering is performed on resultant dataset. Analysis is performed on
the cluster formed and list of possible criminals are generated.

IV. ALGORITHM USED

K-means Clustering
K-means clustering algorithm proceeds by partitioning n objects into k clusters. Each object belongs
to the cluster with the nearest mean. This method produces k different clusters with greatest possible

@IJRTER-2018, All Rights Reserved 34

 International Journal of Recent Trends in Engineering & Research (IJRTER)
Volume 04, Issue 02; February- 2018 [ISSN: 2455-1457]

distinctions. K-Means clustering aims to minimize total intra-cluster variance, or, the squared error
function.
𝑗
J = ∑𝑘𝑗=1 ∑𝑛𝑖=1 ||𝑥𝑖 − 𝑐𝑗 ||^2

Where J = objective function

k = number of clusters
n = number of cases
𝑥𝑖 = case i
𝑐𝑗 = centroid for cluster j

Algorithm
1. Define the value of k and cluster the data into k groups.
2. Select k points at random as cluster centers.
3. Calculate the Euclidean distance function and assign objects to their closest cluster center.
4. In each cluster calculate the centroid or mean of all objects.
5. Repeat steps 2, 3 and 4 until each cluster are assigned the same points in consecutive rounds.

V. CONCLUSION
The number of crimes have increased as many people do not register FIR thinking that their
confidentiality will be hampered and also a lot of time will be consumed. A common vision is required
to ensure security and encourage people to report misconducts, thereby preventing crimes in the India.
The time has come to prioritize solving crimes in India at a faster rate.

REFERENCES
I. Archana Iyer, Prachi Kathale, Sagar Gathoo, Nikhil Surpam “EPolice System- Fir Registration and Tracking
Through Android Application”, International Research Journal of Engineering and Technology (IRJET), Volume:
03 Issue: 02 | Feb-2016.
II. Saurabh Zadikar, Jaan Momin, Datta Mule, Avinash Chavan “Criminal Investigation and Identification System
(CI2S)”, IJCSMC, Vol. 4, Issue. 2, February 2015.
III. Lalitha Saroja Thota, Mohrah Alalyan, AL-Otaibi Awatif Khalid, Fabiha Fathima, “Cluster Based Zoning of
Crime Info”, 978-1-5090-5814-3/17/$31.00 ©2017 IEEE.
IV. Malathi. A, Dr. S. Santhosh Baboo, “An Enhanced Algorithm to Predict a Crime Using Data Mining”,
International Journal of Computer Applications” (0975 – 8887) Volume 21– No.1, May 2011.
V. Sagar Bhise,Prof. Sweta Kale, “Efficient Algorithm to find Frequent Itemset Using Data Mining,”International
Research Journal of Engineering and Technology (IRJET), Volume: 04 Issue: 06 | June-2017.
VI. Shekhar R. Gaddam, Vir V. Phoha, Kiran S. Ballagani, “ K-Means+ID3: A Novel Method for Supervised
Anomaly Detection by Cascading K-Means Clustering and ID3 Decision Tree Learning Methods,” IEEE
Transaction on Knowledge & Data Engineering, vol.19,no. , pp. 345-354, 2007.44.
VII. Muhammad Baqer Mollah, Sikder Sunbeam Islam, [Link] Ullah, “Proposed E-Police System For Enhanced
E-government Services of Bangladesh,”IEEE/OSAIIAPR International Conference on Informatics, Electronics
and Vision, Vol. 8, Issue. 6, November 2011.
VIII. Hongjian Qiu, Yihua Huang, Rong Gu, Chunfeng Yuan, “YAFIM: A Parallel Frequent Itemset Mining Algorithm
with Spark,” 2014 IEEE 28th International Parallel & Distributed Processing Symposium Workshops.

@IJRTER-2018, All Rights Reserved 35