BALLON

An Ontology for Forensic Ballistics Domain

Arif Yılmaz, Hilal Tarakçı and Serdar Arslan

TUBITAK Space Technologies Research Institute, Ankara, Turkey

Keywords: Semantics, Forensic Science, Ballistics, Ontology Development.

Abstract: Forensic ballistics is the domain that analyzes the firearm usage in crimes, thus assisting in exposing

connections between crime scenes. While concept of the ballistics domain is strict and well defined; to the

best of our knowledge, there is no standard representation of knowledge on this domain open to the public

use. In our study, we represent an open ballistics domain ontology in a ballistics identification system and

our aim is to acquire an effective reasoning capability. The proposed ontology, models the real world

relationships between the concepts, thus forming a very close semantic representation of the ballistics

domain. Therefore, in our ontology, reasoning capability is effectively used in order to set up relationships

between concepts automatically.

1 INTRODUCTION

Forensic ballistics is the domain that analyzes the

firearm usage in crimes, thus assisting in exposing

connections between crime scenes. This science

depends on the fact that each firearm leaves unique

marks on the bullet and cartridge case it fires due to

the manufacturing imperfections on the firearm.

Therefore, these marks on cartridge cases and bullets

are compared to each other in order to identify the

ones which were fired from the same firearm.

The comparison of bullets and cartridge cases is

a complicated task. In order to automate this

process, some ballistics identification systems are

developed (Condor Homepage, 2010) (Baldur,

2001). Balistika series automated firearms

identification system (Leloglu et al., 2003a and

2003b) is one of these systems, which enables taking

images of bullets and cartridge cases via a camera,

comparing the image against a huge database of

previously taken images according to some given

criteria in a fast and efficient way. Besides, roles of

the people interfering in the crimes will also be

gathered in order to support related queries.

While the concept of the ballistics domain is

strict and well defined; to the best of our knowledge,

there is no standard representation of knowledge on

this domain open to the public use. This situation

results in incompatibility between different ballistics

identification systems. Recently, there is an ongoing

project (Yates et al., 2009) claiming to address this

incompatibility problem by using ontologies at

several layers in their project stack. According to

their paper, the ontologies being developed in this

project have not been disclosed yet. Besides, the

study is only an abstraction of the application

concepts at a high level. Thus, it does not address

the issues related to reasoning power of ontology

In this study, an open ballistics identification

domain ontology is presented. It aims to add

superior reasoning capability to new generation

Balistika System: BALISTIKA 2010. The strictness

and well defined nature of forensic ballistics domain

makes this goal achievable. The proposed ontology

defines the real world relationships between the

concepts, thus forming a very close semantic

representation of the domain. Therefore, reasoning

capability is effectively used in order to set up

relationships between concepts automatically.

In Section 2, some background information and

related works are given. Our ontology is explained

with a test scenario in Section 3. Finally,

conclusions with further research directions are

presented in Section 4.

2 BACKGROUND

The evidences (“cartridge cases” and/or “bullets”)

collected from a crime scene are entered in the

392

Yilmaz A., Tarakçı H. and Arslan S..

BALLON - An Ontology for Forensic Ballistics Domain.

DOI: 10.5220/0003095603920395

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2010), pages 392-395

ISBN: 978-989-8425-29-4

 2010 SCITEPRESS (Science and Technology Publications, Lda.)

system as an envelope termed as “a case”. When “an

evidence” is found in the crime scene, the high

resolution images of this evidence or information

derived from these images are compared with the

other images of the evidences in the system

database. If “a match” is found in the database, the

evidences, groups of those evidences and wrapper

cases are linked in the same chain.

The relationship between cases is a crucial goal

of the ballistic investigation. This relationship

defined as a “named chain”, can be helpful in

investigation of other cases in this chain; so that if a

firearm is found or a person is identified in one of

these cases, this information can be used in the

investigation of the current case. Besides; whenever

evidences from each chain are matched in a latter

comparison process, these chains have to be merged

since all evidences linked within the same chain are

shot from the same firearm.

A domain ontology is a reusable vocabulary of

concepts and their meanings, the relationships

between these concepts, and activities in a particular

domain enabling reasoning capability within that

domain (Gómez-Pérez et al., 2004).

In (Brinson et al.,2006), a cyber forensic

ontology is proposed in order to use in curriculum

development and educational materials. However,

the ontology is not specialized for ballistics domain

and not suitable for machine process purposes.

Fortunately, this ontology is appropriate as an upper

ontology for our domain of concern as future work.

To the best of our knowledge, there is currently

only one study proposing an ontology about forensic

ballistics domain. (Yates et al., 2009) tries to solve

this problem by the development of interoperable

data and systems as part of Odyssey Project

(Odyssey Project Homepage, 2010).

3 BALLON: BALISTIKA

2010 ONTOLOGY

Fortunately, ballistics identification is a strict

domain and has well defined nature. This feature

facilitates describing it in a formal way. The entities

in real world ballistics domain and the ones

constituting the ontology conforms each other

making understanding of the ontology easier.

3.1 Ballistics Ontology

In BALLON, case concept is defined to be an

envelope that wraps all the ballistics evidence

discovered at the crime scene besides metadata

about the crime scene such as the region, the people

and roles they played in the committed crime etc.

The cartridge cases and bullets found are separated,

since comparison of a cartridge case with a bullet is

meaningless; in other words, cartridge cases are

compared against cartridge cases, bullets are

compared against bullets

In a crime scene, the possibility of finding sister

ballistics evidences is very high, since most likely

several shots made by the same firearm. Therefore,

the forensic ballistics expert analyzes the evidences

and groups the sister evidences manually, forming

evidence groups in order to increase the efficiency

of the automated ballistics identification system. In

the ontology there are two subclasses of evidence

group: cartridge case group and bullet group.

Similarly evidence has two subclasses; cartridge

case and bullet. In general, this situation holds for

almost all concepts which are subject to this

cartridge case-bullet separation.

Whenever a firearm is discovered in a crime

scene, it is associated with the case. The test shoots

for the firearm are carried out and cartridge cases

and bullets obtained during test shoots are called as

test evidences of that firearm (test cartridge cases

and test bullets). In the ontology, test evidence

groups are evidence groups that have test evidences

shot from the same firearm. In other words, if an

evidence group is defined in the ontology and some

test evidences are indicated to be in that evidence

group, the evidence group automatically becomes a

test evidence group.

The link established between evidences,

evidence groups or cases forms a chain. The

concepts that are claimed to be in the same named

chain are related to the same firearm somehow.

Whenever two cartridge cases or bullets are

indicated to be sisters, the ontology asserts that all

the evidences that are in the same evidence group

with those two evidences are in the same chain.

Moreover, the cases those evidence groups belong to

are also asserted to be linked with that chain, as

well. If the test evidence of a firearm is indicated to

be in that chain, other cases in that chain are also

asserted to be related to that firearm.

Besides ballistics evidence, cases are also related

to people and roles that people play in the case are

indicated. In BALLON, people are automatically

classified as suspect, victim or witness based on

their role in the cases. For instance, if the person has

a grabber or murderer role in a case, he/she is

classified as the suspect of that case.

BALLON - An Ontology for Forensic Ballistics Domain

393

3.2 A Test Scenario

The aim of the ballistic investigation is to find the

interfered people and/or firearms of the case being

examined. For this purpose, the evidences gathered

from the crime scene must be analyzed efficiently.

The firearms, cartridge cases and bullets can be used

for the analyzing process as well as people interfered

in the case.

An example scenario (shown in Figure 1) is as

follows; Cases are represented as rectangles and

evidences (black circle) of the same firearm are

grouped together into evidence group which is

shown as gray circle in the figure.

The Police were informed a murder in Region1

(Case 1). Three cartridge cases have been found in

crime scene while there was no suspect caught. The

ballistics expert has stated that these cartridge cases

are ejected from the same firearm which has not

captured in this case. These evidences have been

registered into the identification system in order to

find the suspect(s) and firearm if they have been

already found in previously registered cases.

Unfortunately, the system has returned no match for

those evidences.

Several months later, another crime was declared

in the same region (Case 2). This time, there was a

firearm found in crime scene, while there were two

cartridge cases. Test shots of found firearm have

been taken in order to compare to other evidences

previously stored in the system but no match found

for that firearm. Meanwhile, the expert has decided

that two cartridge cases found in the crime scene

have not been ejected from that firearm. So, they

were registered into the system to find a match. The

system has offered a match with evidence from Case

1. When the expert approved this match, a link

between those evidences has been set up (Chain A).

Our work, automatically delegates this link to

groups of evidences (shown as gray circle) and also

to the wrapper cases.

In another region (Region 2) similar cases (Case

3 and Case 4) have occurred and this time, the chain

(Chain B) has been set up between them which has a

firearm found. Later, the evidences from Case 2 and

Case 3 have been declared as sisters and while these

different evidences belong to different chains, these

chains have to be merged (dotted arrow in the

figure); since a chain is a container of evidences

shot from the same firearm. An evidence cannot

belong to two different chains because an evidence

cannot be shot from two firearms.

With the help of inference capability on our

ontology, this relationship is automatically set up

and the chains A and B are merged. This merge is

carried out by classifying two chains as identical.

Without this reasoning capability, plenty of code

must be written in order to carry out this operation.

Fortunately, web ontology language (OWL

Homepage, 2010) enables us forming automatically

defined classes, defining necessary and sufficient

conditions to fall into these classes via restrictions.

In our ontology, there are automatically defined

classes for evidences, evidence groups, cases and

firearms for each named chains.

Figure 1: An example scenario for querying the ontology.

If a person is declared to be the owner of the found

firearm in Case 4 and the firearm is linked to some

other cases via chain, that person becomes the

suspect of other cases automatically. Moreover,

people interfered in the case with some role are also

subject to reasoning. As a result of reasoning

capability, people are classified according to their

role. In the ontology, there are several roles and each

person classification is a defined class that gathers

people with roles specified in its necessary and

sufficient condition. For example, the suspect of

Case1 is reasoned as murderer in Case 4 and as

grabber in Case 2 automatically depending on the

type of the cases.

KEOD 2010 - International Conference on Knowledge Engineering and Ontology Development

394

3.3 Implementation and Querying

The proposed ontology is created by using Protégé

(Protégé Homepage, 2010) which is a well-known

powerful ontology editor and knowledge acquisition

system. Afterwards, the aforementioned test scenario

is implemented by loading the previously created

ontology, feeding it with required individuals for

that scenario and serializing the final ontology with

the help of Jena (Jena Homepage, 2010) which is a

java framework for building semantic web

applications. The consistency of the produced

ontology is checked via using Pellet (Pellet

Homepage, 2010) reasoner which run seamlessly in

Protégé.

Besides the basic test scenario, the representation

and reasoning power of ontology is also tested

against some fundamental queries. A query module

is implemented in Java using ontology API provided

by Jena in order to accomplish these queries and

some of them listed below. In fact, this query

module can easily be extended to support more

complex queries which are combinations of the

listed fundamental ones. However, in the scope of

this study, the focus is on the proposed ontology and

its reasoning capability.

Some of the fundamental queries and how they

are accomplished are as follows:

1. Query: “Show Named Chains owned by the

Case C1”.

o Result: “ChainACase” ,“ChainBCase”.

2. Query: “Show Cases of Named Chain A”.

o Result: “C1”,”C2”,”C3”,”C4”.

3. Query: “Show identical Named Chains of

Named Chain A”.

o Result: “Chain A”, “Chain B”.

4. Query: “Show Cases interfered by Person 4”.

o Result: “C3”.

5. Query: “Show the same Cases of Named Chain

A and Named Chain B”.

o Result: “C1”,”C2”,”C3”,”C4”.

4 CONCLUSIONS

The lack of a common and open unified formal data

model in ballistics domain results in incompatibility

amongst existing ballistics identification systems.

However, to the best of our knowledge, an open

ballistics ontology is not proposed yet. In our study,

we represent an open domain ontology for a

ballistics identification system. Our ontology claims

to be a formal representation of the domain,

covering required key concepts and promising an

effective reasoning capability.

As a future study, we will try to integrate our

ontology with the relational database used in

BALISTIKA 2010 project and enhance query

support for practical use of proposed ontology.

ACKNOWLEDGEMENTS

This work is supported by TÜBİTAK 107G194

Research Grant.

REFERENCES

Condor Homepage, 2010, http://www.biscondor.com.

Baldur R., 2001, Inventor, Forensic Technology WAI Inc.,

Assignee, Fired cartridge case examination method

and imaging apparatus, US Reissued Patent RE37,392

E .

Leloglu, U. M, Sakarya, U., Tek, B., Cilingir, O. &

Yerlikaya, M. 2003a, ‘A new automated firearms

identification system using 3D information-Balistika,’

in Forensic Science Int., vol. 175, no. 2-3 , pp. 209-

217.

Leloglu, U. M, Sakarya, U., Tek, B., Cilingir, O. &,

Tunali, E. 2003b, ‘A comparative performance test of

a new ballistic analysis system-BALISTIKA,’ in

Forensic Science Int., vol. 1 no. 136 pp. 163-164.

Gòmez-Pèrez, A. Fern´andez-L´opez, M. & Corcho, O.

2004, Ontological Engineering, 3rd edn. Springer-

Verlag.

Brinson, A, Robinson, A. & Rogers, M. 2006, ‘A cyber

forensics ontology: Creating a new approach to

studying cyber forensics’, 2006. [On-line]. Available:

http://www.dfrws.org/2006/proceedings/5-

Brinson.pdf. [Accessed June, 2010].

Yates, S., Jopek, L., Mitchell, S. J. & Wilson, R. 2009,

‘Semantic Interoperability Between Ballistic Systems

Through the Application of Ontology,’ in Proceedings

IADIS International Conference WWW/Internet, Italy,

pp. 153-157.

Odyssey Homepage, 2010, http://odyssey-project.eu/

OWL Homepage, 2010, http://www.w3.org/TR/owl-

features/

Protege Homepage, 2010, http://protege.stanford.edu

Jena Homepage, 2010, http://jena.sourceforge.net

Pellet Homepage, 2010, http://clarkparsia.com/pellet/

BALLON - An Ontology for Forensic Ballistics Domain

395