AN INTEGRATED E-RECRUITMENT SYSTEM FOR CV
RANKING BASED ON AHP
Evanthia Faliagka, Konstantinos Ramantas, Athanasios Tsakalidis, Manolis Viennas
Computer Engineering and Informatics Department, University of Patras, Patras, Greece
Eleanna Kafeza
Department of Maekting & Communication, Athens University of Economics & Business, Athens, Greece
Giannis Tzimas
Department of Applied Informatics in Management & Finance, Faculty of Management and Economics
Technological Educational Institute of Messolonghi, Messolonghi, Greece
Keywords: e-Recruitment, Knowledge management systems, Recommendation systems, Analytic hierarchy process.
Abstract: In the last decades the explosion of Information and Communication Technologies has led to a whole new
scenario concerning peoples’ accessibility to new job opportunities and companies’ options for employing
the right person for the right job. But, is there a way to exploit today’s technological advances as well as
people’s web presence in order to achieve this goal? In this work we present a set of techniques that makes
the whole recruitment process more effective. We have implemented a system that models the candidate’s
CVs in HR-XML, and ranks the candidates based on AHP (Analytic Hierarchy Process). Finally, it presents
the results to the recruiter who evaluates the top candidates and takes the final decision.
1 INTRODUCTION
The rapid development of modern Information and
Communication technologies (ICTs) in the past few
years and their introduction into people’s daily lives
has led to new circumstances at all levels of their
social environment (work, interpersonal relations,
entertainment, etc). People have been steadily
turning to the web for job seeking and career
development, using web 2.0 services like LinkedIn
and job search sites (Bizer, 2005). On the other
hand, a lot of companies use online knowledge
management systems to hire employees, exploiting
the advantages of the World Wide Web. These are
termed e-recruitment systems and automate the
process of publishing positions and receiving CVs.
The online recruitment problem is two-sided: It
can be seeker-oriented or company-oriented. In the
first case, the system recommends to the candidate a
list of job positions that better fit his profile. In the
second case recruiters publish the specifications of
available job positions, and the candidates can apply,
submitting their CVs.
Many approaches can be applied to automate the
e-recruitment process combining techniques from
classical IR (Kessler, 2009). These include
collaborative filtering techniques (Rafter, 2000),
relevance feedback (Kessler, 2009), semantic
matching (Mochol, 2007), multi-agent systems (
De
Meo
, 2007) etc. Their main drawback comes from
the fact that the CVs in these works are either
submitted by the user in an arbitrary format or are
mined automatically from the Web or other sources
(i.e. from server logs).
In this work we have implemented an integrated
company oriented e-recruitment system that
automates the candidate evaluation. Our approach
differs from conventional e-recruitment systems in
that we don’t accept CVs in a document format, but
rather mandate that applicants fill-in predefined web
forms. Additionally, it models the candidates’ CVs
in HR-XML representation and subsequently
provides a ranking of the applicants, scoring their
147
Faliagka E., Ramantas K., Tsakalidis A., Viennas M., Kafeza E. and Tzimas G..
AN INTEGRATED E-RECRUITMENT SYSTEM FOR CV RANKING BASED ON AHP.
DOI: 10.5220/0003337901470150
In Proceedings of the 7th International Conference on Web Information Systems and Technologies (WEBIST-2011), pages 147-150
ISBN: 978-989-8425-51-5
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

qualifications for the given position requirements.
The scoring and ranking process is based on
Analytic Hierarchy Process, or AHP (Saaty, 1990).
2 CV SUBMISSION AND
MODELING
In on-line recruitment systems, candidates typically
upload their CVs in the form of a document with a
loose structure, which must be considered by an
expert recruiter. This incorporates a great asymmetry
of resources required from candidates and recruiters,
resulting in candidates uploading the same CV in
numerous HR agencies that become overwhelmed
with thousands of CVs. In this work, we follow a
different approach in the CV submission process,
which is detailed in this section, along with the CV
modelling in HR-XML format.
2.1 CV Submission
In the proposed system, we mandate that applicants
submit their CVs in a structured way, filling-in
predefined web forms. These web forms include
many closed-form questions that examine the
candidate’s professional qualifications and his
personality and aptitudes. There are also open-type
questions to be considered by human recruiters. The
forms designed are divided in the 4 sections.
In the first section, which is the education and
qualification section, the candidate fills in his
academic degrees (BSc, MSc, PhD) and professional
qualifications. The candidate is expected to be able
to prove all entered information in this section.
In the second section, the experience section,
there are questions about the applicant’s professional
history. These include his years of experience, the
candidate’s loyalty, his former position titles and the
organizational culture of his previous jobs.
In the personality section, the candidate is asked
to perform a self-assessment of his personality. The
personality traits are divided in four broad
categories, as shown in Figure 1. From these
answers an average score is calculated for each
category. We plan to enrich our system with online
psychometric tests that will give us a more accurate
picture of the candidate’s personality.
In the last section we give the opportunity to the
candidate to write about his competencies. The
candidate could report being good in numbers,
having writing skills, social skills, or scientific /
analytical thinking. In this way, we can give an
opportunity to “unproven” juniors with talents and
potential to build their careers.
Figure 1: Self-assessment of candidate’s personality.
2.2 CV Model
In the proposed system the CVs entered by the
applicants, following the CV submission process,
are represented in HR-XML format. HR-XML is a
library of XML schemas that supports a variety of
business processes related to human resource
management and was developed by the HR-XML
Consortium. It includes schemas to represent all the
necessary information about a candidate.
Representing the CVs in HR-XML allows HR
agencies and companies to exchange CVs in a
machine readable, standardized format which is easy
to process, automating part of the recruitment
process. Our system allows the candidate to
download the XML representation of his submitted
CV, which he can then re-submit to another
compatible system avoiding manual re-entry.
3 RECRUITMENT PROCESS
In this section, we will present the recruitment
process followed in the proposed system. As seen in
Figure 2, the process starts with the candidates
submitting their CVs in the system’s web interface.
These are formatted in HR-XML representation, and
stored in the system’s XML-enabled database. This
allows preserving the structure of the CV as an XML
document.
When a position opens, the recruiter follows a 3-
stage online recruitment process. These stages
include a pre-screening of unqualified candidates, an
automatic online background search and finally the
ranking of candidates. In what follows, we present in
detail the implementation of the system modules.
3.1 Filtering Module
The filtering module performs an automatic pre-
screening of candidates, to identify those that meet
WEBIST 2011 - 7th International Conference on Web Information Systems and Technologies
148

Figure 2: The stages of online recruitment.
the minimum requirements of the offered position.
The filtering process is implemented as a series of
XQueries (Boag, 2002). Using XQueries the
recruiter is able to make iterative queries tightening
the minimum requirements. For example, he can
query the system for “candidates with a BSc on
computer science and job experience over 3 years”.
It must be noted that the applicants rejected by the
filtering module are no further considered for
recruitment.
3.2 Internet Activity Screening
An applicant’s internet activity can be relevant for
some positions, as it can testify some skills regarded
critical for a position. Informal activities, like being
contributor to an open source free-time project might
be considered more important than some official
qualifications. It is widely acclaimed that recruiters
perform web searches and take advantage of
applicant’s online activity. We performed Google
searches with the user’s full name, location and
affiliation using the Google AJAX Search API. We
also asked the candidate to voluntarily provide his
personal web site, his blog, the url of an open source
project he contributes, etc. To evaluate the
significance of the submitted information we used
the Google toolbar API and extracted the PageRank
value of each of these pages (Wu 2008).
3.3 Ranking Module
In the ranking stage, the system derives and ranks
the candidates that have passed the filtering stage,
based on classic IR techniques. Each candidate’s
rank acts as a score of how well his profile fits the
recruiter’s specifications. Ranking is based on the
analytic hierarchy process, or AHP.
The first step in the analytic hierarchy process is
to define the ranking criteria and the criteria’s
grades. The grades may take numeric values or take
the form: excellent, very good, average, poor and
very poor and may differ for each criterion. The
criteria we used in our system are:
 Personality traits: the average scores from the
4 personality categories shown in Figure 1.
 Education: The number of years of full-time
formal (academic) learning..
 Work Experience (Years in a related position)
 Loyalty (years the candidate spent per job).
 Internet activity: Rated as described in 3.2
 Skills: The optional skills the candidate fulfils.
The second step is the elicitation of pair wise
comparison judgments. Specifically, the recruiter
has to compare the importance of the
abovementioned criteria, entering weights. These
weights rank the relative significance of each pair of
criteria. For example, the recruiter has to decide how
much more important is work experience from
education. For this procedure a web form was
implemented shown in Figure 3.
From these comparisons a matrix A with
dimensions × is derived, where n is the number
of the criteria. The matrix A is in the form:
A=













…















…





⋮⋮ ⋮









…








(1)
Then we sum each column of the reciprocal
matrix and divide each element with the sum of its
column, normalizing the matrix. The global priority
vector, that stores the weights of the ranking criteria
is obtained by averaging across the rows.
Figure 3:Form with the pair comparisons.
Then we move to the pair wise comparisons of
the candidates with respect to each ranking criterion
forming 9 matrices. The elements of these matrices
are automatically calculated from the candidates’
grades to each criterion, creating 9 vectors that are
the local priority vectors. Then we compute the
overall composite weight, that acts as the candidate
rank, which is the linear combination of the priority
vector computed by the recruiter’s judgments and
the local priority vectors computed automatically.
AN INTEGRATED E-RECRUITMENT SYSTEM FOR CV RANKING BASED ON AHP
149

4 PILOT SCENARIO
In order to demonstrate the system’s functionality a
testing scenario was defined that uses all the
subsystems detailed above. We used as an input the
CVs from 30 graduate students from University of
Patras in Greece.
Two job positions were selected from the liaison
office of the University of Patras. These jobs
required a different set of skills, so that the selection
and ranking of candidates would become apparent.
The first one was for a junior java developer and the
second one for a junior researcher. For the first job
position the prerequisites were java knowledge and
one year of experience, while for the second position
the required qualification was the possession of an
MSc degree. We firstly used the filtering module to
exclude the candidates who didn’t meet the
positions’ prerequisites
At the ranking phase the priority vectors were
calculated as shown in Table 1, where the global and
local priority vectors for the first job position are
shown. We only display the calculations for the first
6 criteria due to space constraints. The second row is
the global priority vector, while the columns
represent the local priority vectors. It is obvious
from the Table 1 that the criterion 6 (the job
experience), has the highest priority with 27% of the
influence.
Table 1: Local and global priorities for the first job
position.
1 2 3 4 5 6
0,08 0,14 0,02 0,05 0,13 0,27
C18 0,09 0,11 0,09 0,08 0,07 0,15
C23 0,06 0,10 0,07 0,09 0,05 0,12
C6 0,09 0,09 0,12 0,07 0,07 0,13
C12 0,08 0,07 0,05 0,06 0,06 0,10
C14 0,07 0,08 0,11 0,05 0,07 0,09
The results of the pilot scenario were very
promising. The top-5 ranked candidates for the two
job positions were different, which is justified by
their different requirements. After evaluating the
skills of the top-5 candidates of the first job position
we verified that they outweighed the others in
technical skills and the experience section, having
participated in open-source projects. At the second
job position the ranking was based mainly on
personality criteria while the experience and the
technical skills were not as important and had
smaller weights.
5 CONCLUSIONS AND FUTURE
WORK
In this work we have proposed and implemented a
company oriented e-recruitment system that assists
the recruiter in his decision-making process. The
applicants submit their CVs in a structured way,
which are represented in HR-XML format. Our
system automatically filters the candidates that don’t
meet the minimum requirements of the offered
position. Finally, the candidates are ranked based on
the Analytic Hierarchy Process. A number of tests
were performed for evaluating the developed
system. We found that the system is able to
effectively match candidates to offered positions
based on their qualifications and competencies.
REFERENCES
Bizer, R. H., Rainer, E., 2005. Impact of Semantic web on
the job recruitment Process. Wirtschaftsinformatik
2005. Physica-Verlag HD.
Kessler R., Béchet N., Torres-Moreno J., Roche M., El-
Bèze M., 2009. Foundations of Intelligent Systems.
Springer Berlin / Heidelberg
Mochol M., Wache H., and Nixon L., 2007. Improving the
Accuracy of Job Search with Semantic Techniques.
Business Information Systems. Springer Berlin /
Heidelberg
De Meo, P. Quattrone, G. Terracina, G. Ursino, D., 2007.
An XML-Based Multiagent System for Supporting
Online Recruitment Services. In Systems, Man and
Cybernetics, Part A: Systems and Humans, IEEE
Transactions. IEEE
Saaty TL, 1990. How to make a decision: The analytic
hierarchy process. European Journal of Operational
Research. Elsevier Science B.V.
Boag S., Chamberlin D., Fernandez M. F., Florescu D.,
Robie J., Simeon J., and Stefanescu M., 2002. XQuery
1.0: An XML query language. W3C working draft
Wu X., Kumar V., Quinlan JR., Ghosh J., Yang Q.,
Motoda H., McLachlan GJ., Ng A., Liu B., Yu PS.,
Zhou Z-H, Steinbach M., Hand DJ., Steinberg D.,
2008. Top 10 algorithms in data mining. Knowledge
and Information Systems. Springer London
WEBIST 2011 - 7th International Conference on Web Information Systems and Technologies
150