Research Article  Open Access
Expert System for Competences Evaluation 360° Feedback Using Fuzzy Logic
Abstract
Performance evaluation (PE) is a process that estimates the employee overall performance during a given period, and it is a common function carried out inside modern companies. PE is important because it is an instrument that encourages employees, organizational areas, and the whole company to have an appropriate behavior and continuous improvement. In addition, PE is useful in decision making about personnel allocation, productivity bonuses, incentives, promotions, disciplinary measures, and dismissals. There are many performance evaluation methods; however, none is universal and common to all companies. This paper proposes an expert performance evaluation system based on a fuzzy logic model, with competences 360° feedback oriented to human behavior. This model uses linguistic labels and adjustable numerical values to represent ambiguous concepts, such as imprecision and subjectivity. The model was validated in the administrative department of a real Mexican manufacturing company, where final results and conclusions show the fuzzy logic method advantages in comparison with traditional 360° performance evaluation methodologies.
1. Introduction
Nowadays, labor competences and competence evaluation represent a real challenge for organizations, which emerged in order to assign the right man to the right job. This evaluation method is based on questionnaires that involve fixed scales with specific values, such as 100%, 75%, 50%, 25%, and 0%. This kind of evaluation reduces the evaluator opportunity to express points of view and causes a rigid evaluation. Fuzzy set theory appears as an important tool to include inaccurate judgments inherent in personnel evaluation process. According to Butkiewicz [1] Fuzzy Logic is a very good tool for decision problems, especially when nonprecise or partially precise description is available. Although it can be applied with success in management problems, fuzzy logic is not common in this area.
Fuzzy Logic is an artificial intelligence (AI) technique [2]. AI comes with the purpose of developing models and programs of the intelligent behavior. One of the approaches of the AI is Logic, with the main objective of formalization of natural reasoning.
Fuzzy logic has two main components: membership functions and fuzzy rules. Using them it is possible to move a qualitative to a quantitative description, for example, to represent linguistic expressions as mathematic expressions. This is very useful when it is necessary to model the expertise of a human expert.
Fuzzy membership functions express the certainty than an element of the universe belongs to a fuzzy set. It represents the degree of truth as an extension of the valuation. Degrees of truth are very often confused with probabilities but they are conceptually different because fuzzy truth represents membership in vaguely defined sets, and not likelihood of an event. These membership functions can take different shapes according to expertise and preferences of the designer.
In these membership functions the axis represents the universe of discourse, and the axis represents the degrees of membership in the interval. Most commonly used functions are triangular, trapezoidal, Gaussian, singleton, Gamma, and so forth. Membership functions can be expressed as a discrete or continuous function. In other words, is a membership function of a set , according to the elements of the universe.
Fuzzy sets are classes of objects with grades of membership. Each set is characterized by a membership function, which assigns a grade of membership to each object based on a characteristic.
When the universe of discourse is continuous and finite, commonly used notation to represent set is where When the universe of discourse is discrete and finite, fuzzy set is commonly represented as For instance, if fuzzy set contains the elements , , , , and with membership degrees of 0, 0.5, 1, 0.5, and 0, respectively, the fuzzy set is expressed as A fuzzy set in a discrete and finite universe of discourse can be represented too as a set of ordered pairs of and its membership degree in as which results in Geometry of fuzzy sets involves three elements: domain, range, and mapping. In this example, geometry of the fuzzy set is domain: , range: , mapping: ; .Graphically, fuzzy set can be expressed as shown in Figure 1.
As in classic logic, fuzzy logic uses three basic operations in fuzzy sets: union, intersection, and complement. However, fuzzy sets have certain characteristics that make them different from classic sets. Fuzzy sets have elements with variable membership degrees, which means that an element of the universe of discourse can belong to one or more fuzzy sets, with different membership degrees.
The first operation on fuzzy sets is intersection. It is the degree of membership that two fuzzy sets share, that is, is the smallest degree of membership of each element in the fuzzy sets. Intersection of two fuzzy sets and is a fuzzy set in the universe of discourse , whose function is given by where represents the intersection of the fuzzy sets and ; represents the minimum operator.
The operation union results as the biggest degree of membership of each element in the fuzzy sets, that is, the highest value of the fuzzy values. Union of two fuzzy sets and is a fuzzy set in the universe of discourse , whose function is given by where represents the intersection of the fuzzy sets and ; represents the maximum operator.
The logic operation complement results as the degree of membership that the fuzzy set needs to reach the unit. The complementary set of a fuzzy set is that whose function is given by Functions that define operations of intersection and union can be generalized using the triangular norm (called norm) and the triangular conorm (called conorm or norm), respectively.
A tnorm operator is a function of two elements that satisfies the following.
Boundary Conditions. This condition implies the generalization of the classic sets:
Monotonicity. This condition implies that a decrease in the degree of membership for the set or will not produce an increase in the degree of membership of the intersection of the sets and :
Commutative Property. This property indicates that the operator is indifferent to the order of the fuzzy sets that are combined:
Associative Property. This property allows calculating the intersection of any number of fuzzy sets, grouped in pairs, regardless of the order of couples:
In the same way, operator conorm (norm) is a function of two elements that satisfies the following.
Boundary Conditions
Monotonicity
Commutative
Associative
Some interesting contributions of fuzzy logic as a technique to model subjective viewpoints are found in [3], where the authors firstly considered decision making problems using fuzzy logic. Probably, the first attempt to apply fuzzy logic to personnel evaluation was proposed in [4, 5]. Another approach can be found in [6]. Cannavacciuolo et al. [4] presented the application of fuzzy set theory to a personnel evaluation procedure. Effectiveness of fuzzy concepts and methods depends on the approach used for the analysis of organizational issues. Fuzzy set theory allows them to model the weak signals existing in evaluation processes and highlights part of the tacit knowledge involved in individual judgments. Usually, researchers, consultants, and managers use a rather qualitative approach to organizational problems. However the natural language is the preferred instrument to describe the organizational conditions because the shades of meaning and the ambiguity of verbal statements allow the company actors to manage diverging opinions, tensions, and conflicts.
These approaches are detailed in [7–11].
On the other hand, the logicalmathematical models tend to represent a world of certainty and coherence where doubts, contradictions, divergences, polysemy, conflicts, and ambiguities are usually typified, dissolved, degraded, and linearized. Within this same conceptual framework, mathematicians, computer scientists, A.I. researchers, and engineers, in search of formal coherence, quantifiable variables, and efficient algorithms, usually tend to use fuzzy set theory without considering complexity and ambiguity in organizational situations, for example, [3, 12–18].
There seems to be a growing trend towards the use of systematic procedures in personnel selection. For instance, Karsak [19] introduced a method that integrates decision makers linguistic assessments about subjective factors such as excellence in oral communication skills, personality, leadership, and quantitative factors such as aptitude test score within multiple objective programming frameworks. The importance level of each goal is considered by applying the composition operator to the membership function of the goal and the membership function corresponding to its fuzzy priority defined by linguistic variables.
Kolarik et al. [20] present an online approach to monitoring human performance in terms of conditional reliability when a task is performed. Unlike traditional human reliability analysis, this approach develops a dynamic model that can cope with constantly changing conditions that affect operator performance. A fuzzy knowledgebased assessment approach is developed in order to deal with uncertainty and subjectivity associated with human performance assessment.
Podofellini et al. [21] assess the influence of the failure of the operators to perform one task on the failure probabilities of subsequent tasks with an approach called technique for human error rate prediction (THERP) and a fuzzy expert system (FES).
Other works include the use of fuzzy logic to evolve an optimal and accurate judgment according to the human thinking model and also to mitigate the commonly occurred biases in human recruitment and selection procedures, as seen in [22].
García et al. [23] propose, through the use of tools based on fuzzy logic, the evaluation of the impact of training in companies, by applying the reasoning characteristic of fuzzy logic, with the aim of complementing and extending the classical logic.
Tosti and Addison [24] refer that a poorly designed 360° feedback system can do more harm than good. The use of commercial 360° software is not always an option due to specific requirements of each organization. In this way, it may be that people who design 360° programs are well versed in assessment and measurement technology and woefully lacking in their understanding of feedback technology. Reliability of the 360 degree feedback is supported by the number and hierarchy of raters, as referred to in [25], assuming that personal qualities are developmental goals. Therefore, software has been designed specifically for this study evaluation.
This paper is organized as follows. Section 1 presents an introduction to the study. Section 2 shows some fundamental concepts about competences, performance evaluation 360° feedback, and competence evaluation methodology; fuzzy logic basis and a proposed fuzzy logic model are shown too. After that, the application of both systems (traditional 360° feedback system and fuzzy logic 360° feedback expert system) at the administrative area of a real manufacturing company in the state of Veracruz, Mexico, is shown. Section 3 shows clear and concise results while discussion about the significance of the results is developed. Finally, Section 4 describes the main conclusions of the study.
2. Expert System for Competences Evaluation 360° Feedback Using Fuzzy Logic
2.1. Competences
Personnel appraisal is considered as performance evaluation, and it is based on formal evaluation programs with reasonable information amount about employees and their job performance.
The literature describes several evaluation methods, each with its own advantages and drawbacks, and there is no ideal or universal method for all people, positions, organizations, and situations. The choice will depend on many other aspects such as(i)position,(ii)characteristics to be measured,(iii)organizational culture,(iv)objectives, achieved or to be achieved,(v)circumstantial elements.
Performance evaluation methods are classified according to the feature they measure characteristics, behaviors, or outcomes, as referred to in [1]. Behavior methods enable the evaluator to identify how far the employee performance is away from a specific scale. These methods describe what actions should be exhibited during the position performance. It is mainly used to provide developmentoriented feedback. According to GomezMejia et al. [26], the main advantage in the performance measure adopting a behaviorbased approach is that criteria or performance standards are concrete. Behavior scales give employees specific behavior examples that can make them successful (or avoid their success) in their work. If an employee knew the required skills for the position and the corresponding aperture in degrees, it could verify, analyze, and control its own behavior according to the requirements.
A competence is an underlying characteristic in the employee related to an effectiveness standard and superior performance in a job or situation, as discussed in [27].
According to LevyLeboyer [28], individual skills and company competences are closely related. Company competences are constituted by the integration and coordination of individual skills; however, these competences require an integration and coordination of knowledge and personal qualities. Individual competences are an individual property. Company competences are developed by individuals, but they belong to the company.
2.2. 360° Performance Evaluation
360° performance evaluation is a sophisticated scheme that allows the employee to be evaluated by its surrounding bosses, peers, and subordinates (see Figure 2). A scheme may include suppliers or customers.
360° performance evaluation can potentially bring a globalized diagnosis about the employee performance, allowing the evaluator to compare different opinions about the level of competence expected in the evaluated person and then to take decisions about how to increase the level of compliance of this competences.
Alles [29] proposes 7 points for 360° evaluation.(1)Identify cardinal and individual competences. If the company has implemented a performance evaluation system, the competences will be the same. Eventually, it is possible to use a reduced number of competences when using the 360° evaluation system.(2)Design the tool. Questionnaires typically constitute the process support (see Figure 3).(3)Select evaluators: superiors, partners, internal customers in other areas, customers, and external suppliers. Customers can be included or not. It is important to emphasize the fact that assessments are anonymous and that evaluators are chosen by the evaluating person.(4)Launch the evaluation process with stakeholders and evaluators.(5)Data processing: most of the time data are processed by external consultants to preserve information confidentiality.(6)Communicate 360° evaluation results to concerned people.(7)The company will receive a consolidated report. This report will be received only by the employee.
In this work, 360 degree feedback methodology proposed by Alles [29] is applied under two different approaches: traditional 360° feedback and fuzzy logic 360° feedback. Likewise, there are some other substantial differences. (1) Data processing is performed by two software applications, each designed for its specific process. Both applications are able to select evaluators randomly and present questionnaires. From this point, there is another difference: (2) first application performs the evaluation in the traditional 360 degree feedback; the second application is an expert system that uses fuzzy logic into the questionnaires to perform evaluation. The third substantial difference, hence, is that (3) expert system does not require a human expert, except when they are designed. No external consultant is necessary anymore, since expert system achieves this objective too.
2.3. Fuzzy Logic Basis
Fuzzy logic is the mapping from an input measurement space to an output measurement space using linguistic variables. It gives the ability to model imprecision by incorporating qualitative components into a quantitative analysis.
Fuzzy logic systems have a narrow relationship with fuzzy logic concepts such as fuzzy sets and linguistic variables. The most popular fuzzy logic systems are Mamdani and TakagiSugeno.
Mamdani fuzzy systems use 4 components (see Figure 4).(i)Fuzzifier: Mamdani system inputs are typically numeric values, coming from some kind of sensor or being results of a process; to be able to operate this value, Mamdani systems translate this value into a special value that can be operated by the inference mechanisms. This translation is done by the fuzzifier, which converts numeric values into fuzzy values that represent the level of pertinence of the different variables of the system to the fuzzy sets.(ii)Fuzzy inference mechanism: once the fuzzifier has translated the fuzzy values, these have to be processed to generate a fuzzy output. Inference mechanism task is to take fuzzy values and generate a fuzzy output based on a fuzzy rules base.(iii)Fuzzy rules base is the way in which Mamdani fuzzy systems have to represent expertise and linguistic knowledge to solve the issue. It is a set of IFTHEN sentences, containing two parts each: antecedent and conclusion. In a Mamdani fuzzy system, antecedent and conclusion are given by linguistic expressions.(iv)Defuzzifier: inference system output is a fuzzy output, so it cannot be interpreted by an external element which only could operate numeric data. To make it possible to operate this data, output is translated to numeric format, and this task is done by the defuzzifier, using one of different procedures such as gravity center or averaged centers.
Fuzzy Logic uses certain essential components to achieve its purpose.
Imprecision. Often the same term is used to describe imprecision and uncertainty in only slightly related areas of measurement. Imprecision in measurement is associated with a lack of knowledge. Imprecision as a probability form is associated with uncertainty about the future event occurrence. Imprecision in description, the imprecision type addressed by fuzzy logic, is connected with intrinsic or builtin imprecision that belongs to the event itself.
Fuzzy logic addresses the issues associated with an intrinsic imprecision rather than those directly concerned with measuring devices failures in the measurements accuracy. Intrinsic imprecision is associated with a phenomenon properties description and not with properties measurement using some external device.
Ambiguity. There are close semantic relationships between the ambiguity idea and fuzziness; in fact, some fuzzy states can be highly ambiguous. Ambiguity connotes the property to have several but plausible and reasonable interpretations. These interpretations can have different belief states. Ambiguity in meaning is a common occurrence in natural languages.
Likelihood and Ambiguity. Fundamentally, the basic confusion between fuzzy logic and probability arises from the idea that they measure the same kind of uncertainty. In strictly semantic, as well as mechanistic, the two forms of uncertainty are different. Propositions in probability address the likelihood of an outcome for some discrete event. The event outcome either happens or does not happen. Propositions in fuzzy logic concern the degree to which an event occurred. While a probability outcome happens unequivocally, a fuzzy event occurrence may involve some degree of ambiguity or uncertainty.
Fuzzy Sets Components. Gregory [30] indicates that the fuzzy logic has two main components: membership functions and fuzzy rules. When using these components it is possible to move the experiences and human preferences from a qualitative description to a quantitative description.
Membership fuzzy functions can take different figures and forms, according the designer experiences and preferences. Typical functions are triangular, trapezoidal, S, Gamma, Gaussian, and exponential. On the other hand, the fuzzy rules are written as IFTHEN couples and reported in tabular form.
The four basic ways in which the fuzzy rules can be achieved are expert experiences and engineering knowledge, human behaviors, models based on a fuzzy system, and learning processes. These methods are not necessarily mutually exclusive.
Membership Functions. In classical set theory, something is completely included or not. This situation can be described by assigning a value of one to all the elements included in the set and the value of zero to the ones not included in it. The function that assigns these values is called “membership function.” The fuzzy sets allow to describe the degree of membership of the object to the concept given by the labels, and allow to assign values between zero and one to the membership function (see Figure 5).
Mathematical Features of Fuzzy Sets. Main characteristics of the fuzzy sets are height, support, cutoff levelα, and nucleus. Height. It is the highest degree of membership of the elements of the set; that is, When the height of a fuzzy set is equal to 1 it is said that it is a normalized fuzzy set. Support. It is the number of elements whose degree of membership is not zero; that is, Cutoff Levelα. It is the set of elements of with a minimum degree α; that is, Nucleus. It is the set of elements of with a degree of membership equal to 1; that is,
Inclusion Functions in Fuzzy Sets. There are standard families for inclusion functions; the most frequent ones are trapezoidal, singleton, triangular, , exponential, and type.
Trapezoidal functions are defined by four points: , , , and (see Figure 6). This function is zero for values lower than “” and higher than “” and one between “” and “” and takes values in range between “” and “” and between “” and “.” It is used in simple fuzzy systems, since it allows defining a fuzzy set with little information and computing membership function values in a simple way.
This function is common for microprocessor based systems since it can be encoded in a similar format as functions, π functions, and triangular and singleton functions (e.g., if points and are combined the result is a triangular function). Trapezoidal function is defined as follows (see (22)):
Trapezoidal functions are suitable to model properties in a range of values, stages, or levels (e.g., young, adult, elder, etc.). Modeling a triangular function can be done through the simplification. For an function and singleton types (but not soft), and transformations can be applied, respectively.
Triangular function () can be defined as indicated in
functions (see Figure 7) are appropriate for modeling properties with an inclusion value different from zero and for a narrow range of values around point .
Linguistic Variables. Linguistic variables take values from natural language, for example, much, little, positive, negative, and so forth. These words are considered as labels within the fuzzy set theory.
Even though linguistic variables aim to assign labels as variable values taken from natural language words, they will be able to assign numerical values too. Then, in the expression “temperature is cold,” the variable “temperature” must be seen as a linguistic variable, since the value cold is assigned as a fuzzy set. However, this variable can also take numerical values such as “temperature is 4°C.”
Fuzzy Rules. Fuzzy rules combine one or more input fuzzy sets, called premises, and associate them with an outcome fuzzy set, called consequence. The fuzzy set premise is associated using AND, OR, and so forth operators.
Defuzzification Process. There are two common methods of defuzzification process: gravity center (centroid) and maximum output. As is shown in Figure 8, both techniques produce different results [31].
(a)
(b)
Both techniques produce reasonable results when they are applied in specific fuzzy models. Gravity center is the most common method, because it combines evidence about rules and response fields are pondered by the total true degree. Gravity center is, essentially, the weighted average of the output membership function.
Centroid Computation. Centroid technique finds the balance point solution in fuzzy zone using the weighted average in the fuzzy region. Arithmetically, the procedure is formulated by where is the th domain value and is the true membership value at this point. Centroid or defuzzification with moments finds a point that represents the fuzzy set gravity center.
2.4. Fuzzy Logic Personnel Evaluation Model
A competence performance evaluation involves subjective viewpoints and evaluator preferences are reflected at the evaluation moment. Very often, evaluators express their perceptions in natural language terms. Nevertheless, as mentioned above, evaluation questionnaires constitute the performance evaluation support, and they are based on punctual values that do not reflect or approximate real viewpoints.
Therefore, it is desirable to have a flexible evaluation tool that facilitates imprecision, ambiguity, and subjectivity handling. In this purpose, fuzzy sets allow a suitable treatment.
The fuzzy logic model proposed is constituted as follows.
Linguistic Variables. Three linguistic variables have been defined, “Scale,” “Frequency,” and “Required Level.”
“Scale” variable refers to the percentage (assigned by the evaluator) indicating how well the employee behavior matches the competence definition.
“Frequency” refers to the percentage obtained when the evaluator answers a question and "rethinks" his/her evaluation determining the number of times the behavior is manifested.
“Required Level” refers the percentage expected by the organization; the individual must cover conduct according the competence definition.
Output Variables. They refer to the fit conduct qualification corresponding to the competence definition given in percentage. Under this consideration there are three cases, “needs to improve,” “satisfies,” and “exceeds.”
Fuzzy Sets Representation. This model uses triangular inclusion functions to represent linguistic variables; the output variable was modeled with trapezoidal functions.
According to the expert, five fuzzy sets define the variable “Scale” possible values (see Figure 9), which are(i)“needs significant development,”(ii)“needs development,”(iii)“competent,”(iv)“highly competent,”(v)“role model.”
Likewise, there are four fuzzy sets to define the variable “Frequency” possible values (see Figure 10), which are(i)“occasionally,”(ii)“half time,”(iii)“frequent,”(iv)“always.”
The “Required Level” variable was modeled through three possible fuzzy sets (see Figure 11), called(i)“low,”(ii)“average,”(iii)“high.”
The output variable has three cases which allow defining fuzzy sets (see Figure 12); these cases are(i)“requires improvement,”(ii)“complies,”(iii)“exceeds.”
Fuzzy Rules. Fuzzy rules must consider all the combinations among input variables sets. Each combination will be associated with an output variable fuzzy set. Sixty fuzzy rules were created in this case, according to the opinion of an expert, who rigorously analyzed each set of inputs, determining the level of performance produced by each rule (see Table 1).
 
I: requires improvement. C: complies. E: exceeds. 
Fuzzification. The fuzzification process includes membership functions calculation for input variables and, then, uses the minimummaximum criterion for variables activation.
Defuzzification. The defuzzification process will allow getting the final fit qualification; the proposed method is the gravity center (centroid) including the following four steps:(1)divide total area in partial areas,(2)calculate partial areas value,(3)calculate each partial area centroid,(4)calculate total centroid.
2.5. Computational Experiments
This section describes the use of the methodology using both traditional 360° feedback and 360° fuzzy logic feedback.
2.5.1. Application of the Expert System in a Real Company
The use of both methods was carried out in the administrative area of a Mexican manufacturing company, located in the state of Veracruz, Mexico, where administrative procedures require staff to be evaluated based on their performance annually. A branch of the organization chart, consisting of five different but related positions was selected to perform the 360° feedback evaluation process (see Figure 13).
2.5.2. Traditional 360° Methodology
(1)Identify cardinal and individual competences. To evaluate a position we will consider five cardinal competences and eight specific competences (see Table 2).(2)Design the tool. This evaluation includes four questionnaires and each questionnaire includes five questions.(3)Select an evaluator. Evaluation includes selfevaluation, boss evaluation, and peers evaluation. To choose evaluators and launch the process we will design and use the software; in this sense, the selection procedure is random.(4)Execute the evaluation process with concerned people and evaluators. As an example, Table 3 shows selfevaluation values for cardinal competences.(5)Data processing: software specifically designed for data processing to obtain the competence level scale value and frequency value was used; then we will average the results for each question (see Table 4). Finally, averages are compared with the Required Level. The same procedure will be used for specific competences treatment.(6)Reports will be delivered only to the evaluated person. Comments and graphs will be printed in two reports: the first one will include cardinal competences and will be given to the company, while the second report will include specific competences and will be given to the employee.(7)Communicate the 360° evaluation results to the concerned people.

(a)  
 
(b)  

(a)  
 
(b)  

2.5.3. Fuzzy Logic Model Implementation
(1) Choose an organizational department.
(2) Choose a post in the selected department.
(3) Choose a required competence in the post.
(4) Select one question in the questionnaire that evaluates the competence.
(5) Evaluator assigns values for input variables “Scale” and “Frequency,” for example,
Scale = 62%
Frequency = 84%
The Required Level assigned by the company = 75%.
(6) Fuzzification process: compute the variable membership values corresponding to the variable Scale:
Compute the variable membership values corresponding to the variable Frequency:
Compute the variable membership values corresponding to the variable Required Level:
The fuzzification process continues with variable activation, performed according to the min.max. criterion (see (28), (29), (30), and (31)); this procedure allows identifying the membership values that will appear in the defuzzification process.
Rule 32 is
Rule 35 is
Rule 44 is
Rule 47 is
The rules 32, 35, 44, and 47 are the activated variables; they have correspondence with a set in the output variable (see Table 5).

Three activated variables are in the “Obey” set; it is necessary to use the min.max. criterion to select one value (see (32)):
The last membership degrees and activated variables are summarized as follows (see Table 6).

(7) Defuzzification process: this procedure includes five steps.
Divide the total area in partial areas (see Figure 14).
Compute the partial area values (see (33)): Compute each partial area centroid (see (34)): Calculate the total centroid (see (35)):
(8) Competences questionnaires have more than one question. In this sense, each questionnaire will have as many fuzzy treatments as questions. At the end of the process, an average qualification for each competence will be obtained.
(9) Evaluation software and evaluation report. Intelligent evaluation systems have questionnaires that emulate human thought since it gives the option to answer questions using linguistic labels (see Figure 15). Once the evaluator has selected a label, it selects a numerical value. Final stage includes an evaluation report,which includes a summarized table with average qualifications for each competence.
3. Results
3.1. 360° Methodology Application
A complete application includes fourposition analysis (see Figure 13); five cardinal competences were developed using the job description (see Table 7). The first position has nine specific competences, while the second position has ten specific competences (see Table 8), the third position has eight competences, and finally the fourth position has ten competences. Required Levels were established with the company.


Applying the complete traditional methodology 360° to the first position, we concluded that the company must improve “work quality” and “contact modalities,” since these competences obtained the major difference with the Required Level (see Table 9).
 
*Competences with major difference. 
The individual report includes specific competences summary.
Applying the complete traditional methodology 360° to the same post, we concluded that an employee must improve “collaboration” and “commitment levelpersonal disciplineproductivity,” since these competences obtained the biggest difference with the Required Level (see Table 10).
 
*Competences with major difference. 
3.2. Fuzzy Logic Model Application
Applying the full fuzzy logic model evaluation to the first position, we concluded that the company must improve “communication” and “contact modalities,” since these competences obtained the lower final qualification.
Employee must focus on “credibility technique” and “commitment levelpersonal disciplineproductivity” (see Figure 16).
4. Discussion
Traditional methodology 360° indicates competences that the company and the employee must care. Fuzzy logic 360° methodology indicates the competences on which the company and the employees must focus their efforts but in addition gives a qualification. This qualification can be arranged and the lowest value will indicate which competences must be immediately attended.
According to Table 11, the traditional 360° and fuzzy logic methodologies conclude that the company must attend the cardinal competence “contact modalities” and the employee must focus on the specific competence “commitment levelpersonal disciplineproductivity.” However, fuzzy logic method indicates that “contact modalities” must be attended after “communication,” and “commitment levelpersonal disciplineproductivity” must be attended after “technical credibility.”

The traditional 360° system involves only two factors, “Scale” and “Frequency,” and these are compared with the third factor “Required Level.” On the other hand, fuzzy logic model can involve three factors like input variables.
Evaluation questionnaires of the traditional 360° methodology are filled with five possible percentages (100, 75, 50, 25, and 0). On the other hand, questionnaires in fuzzy logic expert system have adjustable values scales, allowing improved flexibility for the evaluator.
Thus, the main advantage in fuzzy logic expert system is the human thinking simulation, assigning labels as a qualification, and allowing subjective and ambiguity treatment.
5. Conclusions
Competences performance evaluation 360° is a complete system since it involves different viewpoints to appraise personnel performance. It allows better interpretations, since the evaluation responsibility falls in different evaluators. This kind of evaluation facilitates the competence concept comprehension, and Required Levels are assimilated. Nevertheless, it is a rigid system because the questionnaire filling procedure is strict, since these were designed to assign fixed values.
Fuzzy logic competences evaluation expert system includes complex analysis, due to identification and modeling input and output variables. However, its main advantage is the ambiguity and subjectivity handling, since the evaluator can assign words to stand a qualification. This system is flexible because numerical adjustable values can be assigned to behaviors. Graphical interpretation helps to obtain suitable feedback. Even more, final processing reports can be obtained easier and faster. Therefore, it represents an excellent tool for competences monitoring, given the importance that the staff appraisal process has for human resource management, in the areas of recruitment and selection, job evaluation, identification of training needs, and so forth, and the value that results from having nonsubjective and biasfree assessments.
The application of an expert system to performance evaluation in a Mexican manufacturing company allows knowing its effectiveness against traditional techniques.
This work brings innovative contributions to soft computing and human resources management solutions, finding new ways to apply artificial intelligence techniques by means of computer applications to processes that typically were performed by humans.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
References
 B. S. Butkiewicz, “Selection of staff for enterprise using fuzzy logic,” in IEEE International Conference on Systems, Man and Cybernetics, vol. 4, Warsaw University of Technology, Warsaw, Poland, 2002. View at: Publisher Site  Google Scholar
 E. Gómez, Modelo difuso de planeación agregada [Doctoral thesis], Centro de Ingeniería y Desarrollo Industrial, Querétaro, México, 2009.
 L. A. Zadeh, “Making computers think like people,” IEEE Spectrum, vol. 21, no. 8, pp. 26–32, 1984. View at: Google Scholar
 A. Cannavacciuolo, G. Capaldo, A. Ventre, and G. Zollo, “Linking the fuzzy set theory to organizational routines. A study in personnel evaluation in a large company,” in Proceedings of the 2nd IEEE International Conference on Fuzzy Systems, pp. 667–672, San Francisco, Calif, USA, April 1993. View at: Google Scholar
 A. Cannavacciuolo, G. Capaldo, A. Ventre, and G. Zollo, “An approach to the evaluation of human resourses by using fuzzy set theory,” in Proceedings of the 3rd IEEE Conference on Fuzzy Systems, Orlando, Fla, USA, 1994. View at: Google Scholar
 J. Stahl, “A method for personnel selection in concurrent engineering using fuzzy sets,” in Fuzzy Sets in Engineering Design and Configuration, H.J. Sebastian and E. K. Antonsson, Eds., Kluwer Academic Publishers, Boston, Mass, USA, 1996. View at: Google Scholar
 C. Argyris and D. A. Schon, Organizational Learning. A Theory of Action Perspective, AddisonWesley, Reading, Mass, USA, 1978.
 K. S. Cameron, “Effectiveness as paradox: consensus and conflict in conceptions of organizational effectiveness,” Journal of Management Science, vol. 32, no. 5, pp. 539–553, 1986. View at: Google Scholar
 D. J. Orton and K. E. Weik, “Loosely coupled system: a reconceptualization,” Academy of Management Review, vol. 15, no. 2, pp. 203–223, 1990. View at: Google Scholar
 C. R. Schwenk, “Cognitive simplification processes in strategic decisionmaking,” Strategic Management Journal, vol. 5, pp. 111–128, 1984. View at: Google Scholar
 K. E. Weick, “Educational organizations as loosely coupled systems,” Administrative Science Quarterly, vol. 21, pp. 1–19, 1976. View at: Google Scholar
 C. Carisson, “On the relevance of fuzzy sets in management science methodology,” in Fuzzy Sets: Theory and Applications to Policy Analysis and information System, P. P. Wang and S. K. Chang, Eds., Plenum Press, New York, NY, USA, 1980. View at: Google Scholar
 B. R. Gaines, L. A. Zadeh, and H. Zimmermann, “Fuzzy sets and decision analysis—a perspective,” in Fuzzy sets and Decision Analysis, H. J. Zimmermann and etal, Eds., NorthHolland, Amsterdam, Netherlands, 1984. View at: Google Scholar  MathSciNet
 R. M. Tong and P. P. Bonissone, “Linguistic solutions to fuzzy decision problems,” in Studies in the Management Science, H. J. Zimmermann, Ed., pp. 323–324, 1984. View at: Google Scholar
 R. R. Yager, “Fuzzy thinking as quick and efficient,” Cybernetica, vol. 23, no. 4, pp. 265–298, 1980. View at: Google Scholar  Zentralblatt MATH  MathSciNet
 P. L. Yu, “Dissolution of fuzziness for better decisionperspective and techniques,” in Fuzzy Sets and Decision Analysis, J. H. Zimmermann et al., Ed., NorthHolland, Amsterdam, The Netherlands, 1984. View at: Google Scholar
 S. French, “Fuzzy decision analysis: some criticism,” in Fuzzy sets and Decision Analysis, H. J. Zimmermann, B. R. Gaines, and L. Zadeh, Eds., Amsterdam, The Netherlands, North Holland Publisher, 1984. View at: Google Scholar
 L. A. Zadeh, “Outline of new approach to the analysis of complex systems and decision processes,” IEEE Transactions on Systems, Man, and Cybernetics, vol. 3, pp. 28–44, 1973. View at: Google Scholar  MathSciNet
 E. E. Karsak, “A fuzzy multiple objective programming approach for personnel selection,” in Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, vol. 3, pp. 2007–2012, Nashville, Tenn, USA, October 2000. View at: Publisher Site  Google Scholar
 W. J. Kolarik, J. C. Woldstad, S. Lu, and H. Lu, “Human performance reliability: online assessment using fuzzy logic,” IIE Transactions, vol. 36, no. 5, pp. 457–467, 2004. View at: Publisher Site  Google Scholar
 L. Podofellini, V. Dang, E. Zio et al., “Using expert models in human reliability analysis—a dependence assessment method based on fuzzy logic,” Risk Analysis, vol. 30, no. 8, pp. 1277–1297, 2010. View at: Publisher Site  Google Scholar
 K. C. Mittal, A. K. Goel, and P. Mohindru, “Fuzzy multicriteria decision making (MCDM) in human resource selection procedure—a case study of Indian IT industry,” BVIMR Management Edge, vol. 6, no. 1, pp. 89–97, 2013. View at: Google Scholar
 R. E. García, G. Felix Benjamin, and R. Bello Perez, “Impact assessment of training with fuzzy logic,” Ingeniare. Revista Chilena de Ingeniería, vol. 22, no. 1, pp. 41–52, 2014. View at: Google Scholar
 D. T. Tosti and R. M. Addison, “360degree feedback: going around in circles?” Performance Improvement, vol. 48, no. 3, pp. 36–39, 2009. View at: Publisher Site  Google Scholar
 R. Hensel, F. Meijers, R. van der Leeden, and J. Kessels, “360 degree feedback: how many raters are needed for reliable ratings on the capacity to develop competences, with personal qualities as developmental goals?” International Journal of Human Resource Management, vol. 21, no. 15, pp. 2813–2830, 2010. View at: Publisher Site  Google Scholar
 L. R. GomezMejia, D. B. Balkin, and R. L. Cardy, Managing Human Resources, Prentice Hall, Saddle River, NJ, USA, 2nd edition, 1998.
 M. Spencer Lyle and M. Spencer Signe, Competence at Work: Models for Superior Performance, 1993.
 C. LevyLeboyer, “Evaluación del personal: los métodos a elegir,” Ediciones Díaz de Santos, 1992. View at: Google Scholar
 M. A. Alles, Gestión por Competencias El diccionario, Granica, Buenos Aires, Argentina, 2nd edition, 2005.
 C. Gregory, Dynamic Economics: Optimization by the Lagrange Method, Oxford University Press, 1997.
 E. Cox, The Fuzzy Systems Handbook: A Practitioner's Guide to Building, Using, and Maintaining Fuzzy Systems, Ap. Professional, 1994.
Copyright
Copyright © 2014 Alberto Alfonso Aguilar Lasserre et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.