![](bg3.png)
Arch - refers to the dactylogram
made up of generally parallel and
convex ridges that run or tend to run
from one side of the print to the other
and very often reveal angular or
vertical ridges. Represented by the
number 1
or the letter A.
Internal Loop: refers to the
dactylogram that presents a delta to
the observer’s right and a nucleus
composed of one or more ridges,
which run from the left of the print
toward the center, recurving and
returning, or tending to return, to the
side from which they originated,
thereby forming one or more loops.
Loops involve the two-way movement of a papillary
line, which must have perfect inflection.
Represented by the number 2
or the letter I.
External Loop: refers to the
dactylogram that reveals a delta to the
observer’s left and a nucleus
composed of one or more ridges that
run from the left of the print toward
the center, recurving and returning, or
tending to return, to the side from
which they originated, thereby
forming one or more loops.
Represented by the number 3 or the
letter E.
Whorl: refers to the dactylogram
characterized by the presence of a
delta to the observer’s left and right
and a varied nucleus, which presents
at least one curved ridge in front of
each delta. Represented by the
number 4 or the letter W.
Accidental: refers to the dactylogram that does
not fit within any of the four primary types cited
before and which is represented by the number 5.
Scar: refers to the dactylogram that presents a
permanent mark caused by a cut, pustule, burn, or
crushing, thereby making its classification within
one of the 5 types cited above impossible and which
is represented by the number 6.
Amputation (or failure): refers to the type in
which a total or partial loss of the phalange is
evidenced, therefore compromising or even
precluding the classification of the primary type, and
which is represented by the number 7.
If we create a fraction in which the numerator is
the number formed by the numbers that represent the
pattern of the fingers of the right hand, extending
from the thumb to the small finger, and the
denominator constitutes the same number for the left
hand, we arrive at the dactyloscopic formula, as it is
known.
Two fingerprints will only be considered
identical when they demonstrate twelve or more
characteristic points having the same configuration
and location. In the majority of countries, these
criteria are required by law for purposes of a positive
identification in criminal cases.
4 CASE STUDY
The purpose of the analysis was to identify the
pattern of the dactyloscopic (fingerprint) profile of
criminals in the Federal District, in comparison to
the national profile, on the basis of the application of
a clustering analysis and statistics, supported by a
clustering model that uses prior knowledge.
It is the task of the National Identification
Institute (INI), a branch of the Federal Police
Department (DPF), linked to the Ministry of Justice,
founded in 1963 and headquartered in Brasilia, to
centralize information and fingerprints associated
with the subjects of police investigations or
individuals charged with crimes within the territorial
boundaries of Brazil, as well as foreign nationals
subject to registration, through the use of the
dactyloscopic identification process. The
Dactyloscopic Research Section has an Individual
Dactyloscopic Archive (AID) comprised by 19
manual archiving machines for individual
dactyloscopic criminals, model NG Class 5500, in
which approximately 1,360,000 records are stored.
Those records have ten fields in which the ten
fingerprints are stored. The archiving of the
individual dactyloscopics is initially accomplished
on the basis of the fundamental types established in
the classification key. The Dactyloscopic Formula
(FD) is the set of numerical symbols representing
the primary classification of the AID.
The database, known as “MECA-Sinic”, was
extracted from the DPF’s mainframe in November
2000 by a domain expert. The database has a total of
502,052 registries. It represents a sample of 37% of
the total number of identification records, randomly
extracted. Complete attribute types: criminal
violation code, sex, skin, birth date, and main types
for each finger. Text attribute types: State. From the
database, all the State’s records matching those of
the Federal District were selected, specifically, a
total of 5,363. The attributes selected for the
clustering analysis were the 10 primary types
corresponding to each finger.
INFORMED K-MEANS: A CLUSTERING PROCESS BIASED BY PRIOR KNOWLEDGE
471