Метод построения правил рекрутингу на основе анализа компетентности специалистов

Lytvyn V.; Vysotska V.; Pukach P.; Bobyk І.; Pakholok B.

Видшеш специфiчнi особливостi задач управлтня квалiфiкованими людськими ресурсами (УКЛР), що iдентифiкують гх як завдання багатокритерiального аналi-зу i прийняття ршень в нечткому серед-овищ^. Запропоновано модель прийняття ршень в задачах УКЛР. Обгрунтовано, що для тдвищення ефективностi та прозоро-стi ршень в УКЛР дощльно застосування багатокритерiальног оптимiзацiг на базi TOPSIS i показат його переваги

Ключовi слова: управлтня контентом, людськ ресурси, рекрутинг, прийняття ршень, нечтке середовище, нечтк мно-жини

□-□

Выделены специфические особенности задач управления квалифицированными человеческими ресурсами (УКЛР), позволяющие идентифицировать их как задачи многокритериального анализа и принятия решений в нечеткой среде. Предложена модель принятия решений в задачах УКЛР. Обосновано, что для повышения эффективности и прозрачности решений в УКЛР целесообразно применение многокритериальной оптимизации на базе TOPSISи показаны его преимущества

Ключевые слова: управление контентом, человеческие ресурсы, рекрутинг, принятия решений, нечеткое среда, нечеткие множества

UDC 004.89

|dOI: 10.15587/1729-4061.2016.85454|

A METHOD FOR CONSTRUCTING RECRUITMENT RULES BASED ON THE ANALYSIS OF A SPECIALISTS COMPETENCES

V. Lytvyn

Doctor of Technical Sciences, Professor* E-mail: [email protected] V. Vysotska PhD, Associate Professor* E-mail: [email protected] P. P u kac h

Doctor of Technical Sciences, Associate Professor** E-mail: [email protected] I. Bo by k PhD, Associate Professor** E-mail: [email protected] B. Pakholok PhD, Associate Professor** E-mail: [email protected] *Department of Information Systems and Networks*** **Department of Mathematics*** Lviv Polytechnic National University S. Bandery str., 12, Lviv, Ukraine, 79013

1. Introduction

Under conditions of active development of innovative information technologies and software development, human resources (IT-professionals) are turning into the main strategic resource of organizations [1], which ensures their long-term competitiveness and achievement of goals, set by the organization [2]. Therefore, development of new conceptual approaches to recruiting IT-professionals is becoming increasingly important and relevant [3]. Recruitment is the process of searching for and selecting personnel for vacant positions in the staff of a company [4]. It is the main function and responsibility of human resources managers and recruiters [1]. Different approaches and information sources are used for recruitment [5]. The main sources are the internal database of a company or an agency, websites for a job search, social capital (or searching for candidates among acquaintances), media, social networks, forums, blogs, etc., the employees of companies-competitors (attracting professionals from other companies), higher educational establishments (inviting young specialists from higher education institutions), and cooperation with recruiting agencies [6].

It is a multifaceted extensive subjective procedure, extended over time and dependent mainly on the level of training of recruiters [7].

2. Literature review and problem statement

As a rule, people who are not qualified or partially qualified in the field and the level of expertise required by the position that the candidates apply for are involved in the recruitment at the initial stages [8]. At the next stages, specialists in subject areas usually take part in job interviews in order to analyse the level of expertise of the applicants who have already been approved after the interview or testing at the initial stages [9]. But they spend a lot of time filtering the information about candidates and their competence [10]. All of this takes a lot of time and resources from companies that offer a job or recruitment companies (it is impossible to have specialists and experts in all possible fields) [11]. It is necessary to automate the partial sub-processes of analysis of expertise of candidates for the positions and to develop approaches to the analysis of the competence level of

©

professionals [12]. Automation greatly reduces the time for selection of skilled workers and decreases the amount of data that need processing; it leads to a decrease in the subjectivity influence regarding recommendations of qualified people to the position. Taking into account the level of competences of applicants during recruiting makes it impossible to hire unqualified professionals [14].

Competences are the dynamic combination of knowledge, understanding, habits, abilities and skills [15]. Development of competences is the aim of training programs [16]. Competences are formed in a variety of academic disciplines and are evaluated at different stages [3].

Competences are divided into three types [2]:

A - Instrumental (cognitive, methodological, technological and linguistic abilities).

B - Interpersonal (communication skills, social interaction and cooperation).

C - System (combination of understanding, receptivity and knowledge, ability to plan changes for the improvement of systems, development of new systems).

In the European dictionary, the term "employability" (qualification for employment) is becoming increasingly popular; it describes the totality of knowledge, skills, habits, a good command of approaches to solving practical situations, as well as the ability and desire of continuous improvement and professional development [2]. Qualification for employment includes the following competences [17]:

- level of self-organization;

- ability to work in a group;

- ability to perform specific tasks;

- communication skills and literacy;

- knowledge of information technologies, etc.

We will note that these are so-called general competences that are independent of the main profile of a chosen profession.

Survey [2], conducted among European employers (mostly representatives of industry and business), demonstrated that the chances of getting a proper place in the labour market depends on:

- skills that characterize employability - 78 %;

- positive attitude to work -72 %;

- appropriate practical experience (internship) - 54 %;

- area of the obtained education and training - 41 %;

- level of academic achievement at a higher educational establishment the applicant graduated from - 28 %;

- name (prestige) of educational institution the applicant graduated from - 8 %.

3. The aim and tasks of the study

The aim of the work is the development of a multi-criteria optimization model of recruitment based on the analysis of competences of an IT-specialist. This will make it possible to automate the process of selection of IT-specialists in the recruitment departments of IT-companies through the development of appropriate software based on the developed model.

To achieve the aim, it was necessary to solve the following tasks:

- development of a formal model of classification of IT-professionals by the level of competence;

- development of the rules of competence evaluation of IT-specialists;

- development of the rules of recommendations for the positions of IT-specialists.

4. A formal model of recruitment based on the analysis of competences of an IT-specialist

Paper [2] contains a prioritized list of 31 competences (K={kj, k2, ..., k31}) according to the programme Tuning (Table 1). The priority is calculated by the method of hierarchy analysis (analytic hierarchy process, AHP) [18]. Each competence belongs to one of the criteria {A, B, C} [2].

Let us sort out the criteria of the Ki competence by categories (Table 1). For each category we will find coefficient of relative importance of criteria of competence wi (as arithmetic mean of priorities of a category) [19]. Using the TOPSIS method [20, 21], we compute coefficients of relative importance of private criteria of competence wj and the weight coefficients of private criteria of competence w^ = w1 *wj (Table 1) [22].

In the course of studying, the profile of distant student S is formed considering each level of competence KA, KB, KC within [0;1]. A student is trained in certain specialty

P = {P1,P2,...,PNp}.

To acquire a specialty, it is necessary to complete several courses

C = {c1,c2,...,cnc} or CP = {Cp1,Cp2,C1P3}, CP c CP c C.

A course consists of disciplines

D = {d^,...^}.

Within each discipline, there are blocks of themes

L = {U2,..,1nl} or LD = {L°,L°,L°}, LDc LD c L,

so

Ki ^ (Di1,Di2,...,Din ) or Ki ^ (Di1,Di2,...,Din )

it is the choice of an expert group, which develops professional educational program. For each Dij, a group of experts

determines weight oij for competence Ki = 1) using

the AHP method [23]. For each Ci we construct a table of relationship between program competences and components of educational programs of a particular specialily (columns contain the list of disciplines, lines contain the list of competences). For example, for specialty 124 "Systems analysis" [24], the expert group defined 48 disciplines with the set of 28 competences and described the relationship between program competences and components of the educational program of a particular specialty. Each cell of this table is the presence or absence of a particular competence in a particular discipline. Accordingly, the matrix of weights of disciplines for the list of competences from the Tuning program is plotted.

S11 S12 ... S1N

Q= °21 °12 ... °2N . (1)

S31,1 S31,2 ... S31,N _

Coefficients of relative importance of criteria received by AHP and TOPSIS

Type Interpretation of criterion Ki Vector of priorities by AHP wi wi Wj = wi *wj

Ability to communicate in another language ki 0,081728 0.30812 0.130369

Ability to communicate orally and in writing in native language ks 0,081728 0.223483 0.094558

Ability to plan time and control it k5 0,073083 0.156996 0.066427

Ability to search for, process and analyze information from different sources k8 0,056095 0.108416 0.045872

A Ability to identify, formulate and solve problems kio 0,046394 0.423112 0.074296 0.031436

Ability to make well- grounded decisions ki2 0,036574 0.050915 0.021543

Knowledge and understanding of subject area and understanding of specialty ki5 0,026993 0.03516 0.014877

Ability to think abstractly, analyze and synthesize ki9 0,015431 0.0247 0.010451

Skills of using information and communication technologies k27 0,005086 0.017915 0.00758

Ability to be critical and self-critical k4 0,073083 0.286411 0.086828

Ability to demonstrate awareness of equal opportunities and gender issues k6 0,064504 0.217059 0.065803

Safety orientation k9 0,04951 0.156114 0.047327

Ability to work in a team ki4 0,028866 0.110389 0.033465

Ability to work in international context ki6 0,024721 0.077361 0.023453

B Ability to act based on ethical considerations ki7 0,020154 0.303159 0.054062 0.016389

Ability to communicate with non-specialists of the same area kis 0,016523 0.037886 0.011485

Interaction and interpersonal skills k2i 0,011812 0.02679 0.008122

Ability to act with social responsibility and civic consciousness k23 0,009635 0.019496 0.00591

Focusing on preservation of environment k28 0,004351 0.014431 0.004375

Ability to learn k2 0,081728 0.334658 0.091605

Ability to produce new ideas (creativity) k7 0,06124 0.262228 0.071779

Ability to apply knowledge in practical situations kii 0,041715 0.206098 0.056414

Ability to conduct research at the appropriate level ki3 0,032705 0.152891 0.04185

Entrepreneur spirit k20 0,013559 0.11133 0.030474

Ability to design and manage projects k22 0,011796 0.080632 0.022071

C Certainty and perseverance in performing received tasks and responsibilities k24 0,008331 0.273726 0.058299 0.015958

Correct understanding and respect for multiculture and differences k25 0,007182 0.042696 0.011687

Ability to work independently k26 0,005859 0.031049 0.008499

Ability to adapt to new situations k29 0,00363 0.023825 0.006522

Ability to evaluate and maintain the quality of completed work k30 0,003158 0.017331 0.004744

Ability to motivate people and to move towards common goals k3i 0,002823 0.013621 0.003728

Because of the lack of professional expert studies of the relationship between program competences from the Tuning program (Table 1) and the components of educational program of a particular specialty, in this paper we assume that

Ojj are equal among themselves and ^gj = 1.

j

Analyzing the obtained weights of disciplines, the expert group estimates the weight of semesters or courses

c^,...,^ (¿ce = 1) at Dj ece.

iНе можете найти то, что вам нужно? Попробуйте сервис подбора литературы.

c j=i

The set R contains specific learning outcomes by the theme blocks

R = (r1,r2,...,rNR) or RL = {ri,r2>».,0 RL c RL c R

or by disciplines Di. G are the learning outcomes by the category of competence at

G = {ga,gb,gc}, g = {

51»g2'...'gNr }'

GA = {g1,g2,...,gm}, GB = {gm+1,gm+2,...,gm+nl,

G = {gm+n+1'gm+n+2'...'gNG}. (3)

Gf c Gf c G, Gf c Gf c G, Gf c Gf c G. (4)

Gf c Ga C G, Gf C Gb C G, Gf C Gc C G. (5)

gd={giD,gf,...,gfD}, gD={gfD+i,gmD+2,...,gmD+nD},

GD = {gmf+nf +1,gmf+nf+2,...,gNGD}. (6)

Fig. 1 displays tree of analysis of relationship between courses, disciplines, themes and learning outcomes.

Rf c Rd c R for discipline Di of students (expert assessment) is assessed by the ECTS scale (Table 3). For a more accurate assessment of a student's level of competence, fuzzy trapezoidal numbers RO are used in the TOPSIS method. In the course of finding trapezoidal fuzzy numbers, correspondence between the ECTS scale and numeric grades was used, which is presented in Table 2.

Next, we find weights of competences for particular student Si

^ = QceR° (7)

and a student's rating by the levels of acquired competences

x(Si)e.

To analyze learning outcomes of student Si, we calculate his level of competence through the calculation of Ri and Gi by the defined set of characteristics aS within [0;100] (Table 3).

Then we normalize RS ^ KS within [0;wi], taking into account the weights of competences (Table 1). For example, KA varies within [0;0.423112] , Kf within [0; 0.303159] and KC within [0; 0.273726]. An analysis of learning outcomes (aS) will be performed with the use of the TOPSIS method and compared with ai from Table 5, which presents the hierarchy of recommended key professions of IT profile and their criteria of importance. Column 3 contains results of calculation by the AHP. The limits of evaluation in case 1 - according to a general education assessment at higher education institutions considering factor of importance of

position (a| 100 = 50 + 50 Pj), and in case 2 - taking into

j=i

account the level of competence. Column "Assessment limits 1" was obtained considering the point scoring assessment of alternatives at higher education institution (that is, a student with total assessment for the entire period of studying, for example, 80,5 points, may qualify for all positions, starting with the 4th, whereas he can not qualify for the first 3 positions; to receive position 13, it is necessary to be assessed within [50; 74.52 87]). Column "Assessment limits 1" was obtained considering factor of importance by AHP (column 3 of Table 4). Column "Assessment limits 2" was obtained without considering significance by AHP (a2 -100 = 50 + (50/15)(i -1)).

Table 2

Linguistic values and corresponding fuzzy trapezoidal numbers

N°. Linguistic values By ECTS Fuzzy values Grade R' Fuzzy trapezoidal numbers Conditional sign

1 bad F 0,15 0-25 (0,0,1,2) 1

2 unsatisfactory FX 0,35 26-49 (1,2,2,3) 2

3 sufficient E 0,55 50-60 (2,3,4,5) 4

4 satisfactory D 0,65 61-70 (4,5,5,6) 6

5 good C 0,75 71-79 (5,6,7,8) 8

6 very good B 0,85 81-87 (7,8,8,9) 9

7 excellent A 1 88-100 (8,9,10,10) 10

Table 4 displays an approximate list of common positions in IT-cluster in Ukraine. Outside Ukraine (for instance, in the EU), this list is much wider. The list was formed and sorted based on the results of a survey of representatives of IT-cluster in Lviv. At present, much to our regret, there are no clearly formulated general requirements to each of the listed positions. Each company determines the range of qualifications for each of the enumerated positions. For example, in some of them, to take up the position of Programmer, it is sufficient to know one of the programming languages, and to take the position of Software Engineer, an applicant must know the basics of algorithmization. The position of Student was introduced to the table conditionally, in order to indicate minimum conditions for making recommendations for

P,

C,

d, d2... d7 d, d2... d7 d, d2... d7

gi G2...Gy ... /T\

ga gb gc ga gb gc

gi g2...gN

.../T\

ga gb gc ga gb gc

gi g2...gN ...

ga gb gc

Fig. 1. Hierarchy of analysis of learning outcomes and the level of competences

C

Table 3

Rules for the formation of competency profile of student

Course Discipline Theme RA RB RC Learning outcomes

Lk RAk Rjk RCk G,jt = + Rj + j

Ci Dj Lk+1 Rijk+1 Rijk+1 Rijk+1 1A b C Gijk+1 = 3(Rijk+1+Rijk+1+Rijk+1)

Lk+2 RAk+2 Rijk+2 R^k+2 1A B C Gijk+2 = 3 (^^+2 + Rijk+2 + Rijk+2)

Learning outcomes GA=¥ Ä GB=1 ?RBk GC=1 » Analysis of results aS*100

further positions. In Ukraine, according to the requirements of Ministry of Education and Science, postgraduate training requires higher qualification than that for other positions in IT-cluster. If recommendations for positions numbered 2-7 allow IT-companies immediately and without any hesitation to hire recommended graduates, the recommendations for positions numbered 8-15 only register the possibility for a graduate to take such positions after appropriate internship and gaining experience at previous positions. Recommendation for position 10, for example, indicates only that a graduate can take positions 2-7 with the prospect of a career growth to position 10 inclusive with the competences acquired in the course of studying. In future he can qualify for higher positions under condition of development of his skills and competences.

Table 4

Coefficients of relative importance of the recommended basic IT-professions

No. Position Coefficient of importance by AHP pi Assessment limits

Case 1 -a1*100 Case 2-a2*100

1 Student 0 50 50

2 Assistant 0.007092 50.3546 53.33333

3 Tester 0.008599 50.78455 56.66667

4 Recruiter 0.011141 51.3416 60

5 Associate Software Engineer 0.014037 52.04345 63.33333

6 Administrator 0.018602 52.97355 66.66667

7 Programmer 0.023866 54.16685 70

8 Software Engineer 0.031904 55.76205 73.33333

9 Senior Software Engineer 0.041297 57.8269 76.66667

10 Team Leader 0.055204 60.5871 80

11 Associate Manager 0.071152 64.1447 83.33333

12 Manager 0.092353 68.76235 86.66667

13 Senior Manager 0.115327 74.5287 90

14 Senior Executive 0.142542 81.6558 93.33333

15 Systems architector 0.170746 90.1931 96.66667

16 Postgraduate 0.196136 99.9999 100

To run an analysis of learning outcomes and to make recommendations, we divide all students into 3 levels of training. In accordance with the value of the competence priority (Table 1), we divide the sorted list of competences into three parts (equal). The first 10 competences (A-5, B-3, C-2) belong to part I, the next 11 competences belong to part II (A-3, B-5, C-3), and the remaining 10 belong to part III (A-1, B-2, C-7). An analysis of the level of competence of a subject according to learning outcomes will be used to make the rules of recommendations for a position or a choice of oc-

cupations (Table 5, 6, b is analysis of competences according to learning outcomes).

Table 5

Rules of determining competence of student S for the analysis of competences according to learning outcomes (b)

level I level II level III

—wA < KA < wA 2 S 1 wA < KA < wA 2 S 1 wA < KA < wA 2 S

0 < K < —wB S 2 —wB < K < wB 2 S —wB < KB < wB 2 S

0 < KC < —wC S 2 0 < KC < —wC S 2 1 wC < KC < wC 2 S

Table 6

Rules of determining recommendations concerning professions of student S for the analysis of competences according to learning outcomes (b)

1 wA < K < 3wA, 2 S 4 1 wA < KA < wA, 2 S —wA < KA <3wA, 2 S 4

Rule 0 < KB <—wB, S 2 0 < KB < — wB, S 2 —wB < KB <3wB, 2 S 4

0 < KC <— wC S 2 0 < KC < —wC S 2 0 < KC < —wC S 2

Position 2 3 4

1 wA < KA < wA 2 S —wA < KA < wA 2 S —wA < KA < wA, 2 S

Rule 0 < KB < — wB S 2 0 < KB < —wB S 2 —wB < KB <3wB, 2 S 4

1 wC < KSC < 3wC 2 S 4 —wC < KC < wC 2 S 0 < KC < —wC S 2

Position 5 6 7

1 wA < K < 3wA, 2 S 4 1 wA < KA < wA, 2 S —wA < KA < wA, 2 S

Rule —wB < K < wB, 2 S —wB < K < wB, 2 S —wB < K <3wB, 2 S 4

0 < KC <— wC S 2 0 < KC < —wC S 2 —wC < KSC < 3wC 2 S 4

Position 8 9 10

1 wA < K < 3wA, 2 S 4 —wA < KA < V, 2 S 4 —wA < KA <3wA, 2 S 4

Rule —wB <KB < wB, 2 S —wB < KB < 3wB, 2 S 4 —wB < KB < wB, 2 S

iНе можете найти то, что вам нужно? Попробуйте сервис подбора литературы.

1 wC < KSC < 3wC 2 S 4 —wC < KC < wC 2 S —wC < KC < wC 2 S

Position 11 12 13

1 wA < KA < wA, 2 S 1 wA < KA < wA, 2 S —wA < KA < wA, 2 S

Rule 1 wB < KB < 3wB, 2 S 4 —wB < KB < wB, 2 S —wB < KB < wB, 2 S

1 1 wC < KC < wC 2 S —wC < KSC < 3wC 2 S 4 —wC < KC < wC 2 S

Position 14 15 16

According to the example of Table 1, KA ranges within [0;wA] ([0; 0.423112]), Kf - within [0;wB] ([0; 0.303159]) and KC - with [0;wC] ([0; 0.273726]).

5. A formal model of IT-specialists classification by the level of competence

The process of making recommendations for a position or classification of IT-specialists by the level of their competence is described by superposition

R = 5° g° P° a, (7)

where sign ° means superposition of functions (output results of function a are the input date for function b, that is, it is a sequence of performing processes from the right to the left); a is the analysis of learning outcomes and compiling the rating by acquired competences; b is the analysis of competences according to learning outcomes; g is the analysis of requirements for a position by the level of competences and learning outcomes; 5 is making recommendations to a student/candidate regarding his profession according to statistics of studying and attainment of competences (5f is compared with 5j(xi); 5f is calculated by learning outcomes of student Si; 5(xi) is the integral index, set within certain limits).

Integral index (coefficient of proximity of compared alternatives) 5(xi), obtained on the basis of evaluation of these private criteria, expresses a certain value of a degree of chance of hiring every candidate xi in the interval [0, 1]. The value of this magnitude allows making the final decision regarding each alternative candidate. In the process of conducting experiments connected with hiring an employee, we analysed the rules for choosing an alternative of recommendation for a position, according to the analysis of requirements for it (Table 7) [11].

A level of competence depends on the learning outcomes over a certain period (for example, 4 years of studying). Each period (year) may have its coefficient of importance according to the expert assessment. As a result of pairwise comparison of the importance of attainment of competences for each of the 4 years of training, the following values of coefficients of importance depending on the year of study were obtained for case I: e1=0,375, e2=0,125, e3=0,125, e4=0,375. Table 8 presents 5 cases (5 different opinions of different groups of experts) of distribution of coefficients of importance of competences for each year of studying out of 4.

As an example, we will analyze learning outcomes of any arbitrary 3 students (x1, x2, x3) over four years for the purpose of their correspondence to 31 competences for the position of Senior Manager (position numbered 13 in Table 4). To do this, with the use of the TOPSIS method, we will calculate trapezoidal fuzzy numbers in Table 9, reflecting one of the equivalents of assessing a student's level of knowledge (Table 2).

According to Table 9, a student can score maximum 12400 points (31*4*100). For example, candidate x1 scored 11160 points (5f = 11160/12400 = 0.9 by general educational requirements for knowledge assessment), candidate x2 scored 9080 (52 = 0.732258), candidate x3 scored 11160 (53 = 0.9). Candidates x1 and x3 have the same number of

points, but there is a difference in their level of training. Candidate x1 studied better in the 1st year, worse - during the remaining years. x3, on the contrary, studied worse at the beginning of training, but had better results in the end. That is why they have different levels of competences according to the chosen specialty. Special subjects are usually taught in senior courses. Accordingly, x1 and x3 have equal rights to claim for a position without taking into account their level of competence in all 31 criteria. Taking into account the competence of years of studying, a matrix of trapezoidal fuzzy numbers is plotted (Table 10). Table 10 also displays aggregated trapezoidal fuzzy numbers. Elements of the matrix of aggregated trapezoidal fuzzy numbers are multiplied by weights of private criteria and results are normalized (Table 11).

Table 7

List of alternatives of recommendations for selection of position

No. Alternative Candidate Explanations concerning hiring

1 8(xi)e [0; 0.25) does not meet requirements for the position candidature is rejected

2 8(xi)e [0.25; 0.50) hardly meets requirements his hiring is a great risk

3 8(xi)e [0.50; 0.65) partially meets requirements for the position hiring is associated with insignificant risk, which can be compensated in the process of working by high indices of other competences

4 8(xi)e [0.65; 0.80) meets requirements for the position some indices can be easily filled in the process of adaptation

5 S(xi)e [0.80; 1.00] fully meets requirements for the position candidature is recommended

Table 8

Coefficients of competence according to year of studying

Pairwise comparison ce Variants of coefficients of importance of competences for each year of studying (cases)

I II III IV V

Year ce 0,375 0.25 0.463184 0.413265 0.568093

Year ce2 0,125 0.25 0.275411 0.292222 0.287771

Year c3 0,125 0.25 0.175972 0.186714 0.10758

Year c4 0,375 0.25 0.085433 0.107799 0.036557

Example of learning outcomes of students x-| x2 x3

No. Criteria Alternatives Expert assessment R' Matrix of trapezoidal fuzzy numbers RO

ce c2 c3 c4 ce c2 c3 c4

1 ki xi A A B B (8,9,10,10) (8,9,10,10) (7,8,8,9) (7,8,8,9)

X2 A B B C (8,9,10,10) (7,8,8,9) (7,8,8,9) (5,6,7,8)