Analytic Hierarchy Process in Production Engineering: A Bibliometry Analysis Analytic Hierarchy Process na Engenharia De Produção: Uma Análise Bibliométrica

— AHP (Analytic Hierarchy Process) tool stands out for its reliability and effectiveness, therefore, the objective of this research was to identify the scientific productions found in the electronic databases chosen by the author on the AHP tool. The date of publication covered by the survey comprises the periods from April 1992 to January 2018. The electronic bases used for the research were Scopus, Web of Science and Scielo, which after filtering only articles that mentioned the method in the title or abstract totaled in 178 articles. The year 2013 was the most productive with 24 articles (13.48%) published. Eight researchers appeared to be more productive because they had more than one published article. The International Journal of Production Research was the most productive periodical (5.62%), China the country with the largest number of publications (34.26%) and four universities had the largest number of articles published (6.74%) and the words that more appeared in article were process, hierarchy, analytic, fuzzy and production. It is concluded that the AHP tool still has great relevance in decision support methods even after 46 years of its creation, and that its potential continues to be promising for the analysis and solution of problems.


INTRODUCTION
The market is in increasing need for quality information, which makes decision support tools a pivotal force for choices involving large risks in companies.
The available tools are the most versatile, providing rational procedures to model problems and represent and quantify variables, taking into account the criteria and weights proposed by the decision makers. Currently, the AHP (Analytic Hierarchy Process) method stands out for its traditionality and reliability.
On the other hand, with the significant increase in the publications, works that classifiy and / or synthesize the knowledge produced become more relevants. Bibliometric, scientometric and web-based studies, as well as systematic reviews, stand out in this line. In this way the reader will have access to a quality material, which describes aspects predominantly quantitative of the state of the art of a certain theme.
When comparing the scientific production of bibliometrics in the Scopus site in 2006 and 2017, which had 205 and 1172 publications respectively, the 471.71% increase in publications empirically affirms the growing relevance of the theme in the academic community.
The present study aims to identify the scientific production about the AHP method pertinent to the Multicriteria Decision Support.

II. ANALYTIC HIERARCHY PROCESS (AHP)
Multicriteria Decision Support (AMD) consists of a set of methods and techniques to assist or support individuals and organizations in making decisions, when habing multiplicity of criteria (GOMES et al. 2002).
Another important step is the choice of the method to be used, which should depend more on its adequacy to the preference structure of decision makers than on the analyst's preference for particular models and methods.
The multi-criteria decision analysis process can be presented with the following steps (GOMES et al. It should also be noted that a family of criteria, that is, the set of criteria used in a given decision situation, must satisfy three conditions (Roy's axioms) to be a coherent family of criteria (ROY; BOUYSSOU, 1993;MELLO et al. 2003): exhaustiveness (it imposes the need to describe the problem taking into account all relevant aspects); cohesion (requires the correct analysis of which are the criteria of maximization and which of the minimization criteria); and non-redundancy (it obliges to exclude criteria that are evaluating characteristics already evaluated by another criterion).
The AHP (Analytic Hierarchy Process) method was developed by Tomas L. Saaty and his first record is dated 1972 in the article An eigenvalue allocation model for prioritization and planning. It is the most widely used and well-known multicriteria method in support of decision-making in negotiated dispute resolution, in problems with multiple criteria (MARINS, SOUZA, BARROS -2009). Ishizaka and Labib (2011) affirm that the structural basis used in the method is inspired by past findings, such as peer comparison rather than weight allocation (Thurstone, 1927;Yokoyama, 1921), the hierarchical formulation of the criteria (Miller, 1966), scale 1-9 based on observational psychology (Fechner, 1860;Stevens, 1957) and the number of items in each level (Miller, 1956).
According to Costa (2002, p. 16-17) this method is based on three stages of analytical thinking: 1 -Hierarchical Construction: Because problems usually have complex resolutions, their structure is made hierarchically to facilitate their understanding. The first level corresponds to the general purpose of the problem, the second level the criteria and the third the alternatives. The number of judgments required to construct a generic judgments matrix A is n (n-1)/2, where n is the number of elements belonging to this matrix. The elements of A are defined by the conditions : • Calculation of global priorities: in this step we wish to identify a global priority vector (PG), which stores the priority associated with each alternative in relation to the main focus.
According to Vaidya and Kumar (2006), one of the main advantages of the AHP method is its flexibility, allowing integration with other techniques for problem solving, such as Quality Function Deployment, Fuzzy Logic and Linear Programming. This enables the decisionmaker to derive the benefits of all the combined methods, helping to achieve the desired results optimally. Tam and Tummala (2001) describe the AHP as being very useful for reaching a consensus when it involves several decision makers with different conflicting objectives.
Tam and Tummala (2001) describes the AHP as being very useful for reaching consensus when it involves several decision makers with different conflicting objectives.
According to Silva and Hamacher (2008), one difficulty that persists in problems structured with the AHP method is the quantification of project quality assessments. Grandzol (2005) says that improper application of the method in unfavorable environments is perceived as oversimplification or as a waste of time.
Although the fault is not blamed on the tool's structure, most decision makers are unaware of the successive improvements that have been developed over time to specific thematic problems (ISHIZAKA, LABIB, 2011). The themes encompassed by the first group are selection, evaluation, cost-benefit analysis, planning and development, priorities and ranking, and decision making. The second group is focused on specific applications in meteorology and medicine related areas, while the third group has AHP problems integrated with the Quality Function Deployment (QFD) tool.
It is also interesting to highlight some works that made use of the method. Kengpol and O'brien (2001) present a decision-making tool for selection of advanced technology, integrating as criteria cost-benefit analysis, an effective decision-making model and a common criteria model for a choice of Time Compression Technologies (TCT).

III. BIBLIOMETRY
Tarapanoff et al. (1995) establish bibliometrics as the study of quantitative aspects of scientific production, sharing and use of information recorded, from mathematical models, to the decision-making process.
According to Lopes et al. (2012), bibliometrics is a quantitative and statistical method to measure indicators of production and dissemination of knowledge, as well as observing the development of several scientific areas and the patterns of authorship, publication and use of research results. The evaluation of the scientific production, important for the analysis of the researchers in the scientific community, is made through the application of several bibliometric indicators, which are divided into indicators of scientific quality, importance and impact.
Bibliometry has laws that use statistical and mathematical techniques that determine the principles of research and ordering in scientific analysis, such as the laws of Lotka, Bradford and Zipf (GUEDES AND BORSCHEVIER, 2005).
According to Urbizagastegui (2008), Lotka's Law states that the number of authors who make n contributions in a given scientific field is approximately 1 ² of those making a single contribution, and that the proportion of those making a single contribution is about 60 percent.
According to Bradford (1934), there is a conception of core formation of journals that address a specific subject. The author states that with the advent of initial articles on a particular topic, they will go through the selective process of journals to which they submitted, and if accepted, will promote the scientific production of articles on the same theme, creating a nucleus of more productive journals followed by areas of less dedication to the theme.
Zipf's Law is based on the principle of least effort: it accounts for the frequency of words in an article. According to Amaral et al. (2004), the law states that if the words appearing in an article are counted and sorted in descending order of number of occurrences, the multiplication of the number of occurrences by the ranking position for each word is a constant.

IV. METHODOLOGICAL PROCEDURES
The article analyzed three electronic databases for this study. Two reference bases (Scopus and Web of Science) and one textual (Scielo). The first two databases are not open to the public, and access has been granted through the partnership established between Brazilian universities and databases, the third being open to the public.
We use the term "Multicriteria Decision Support Tools" OR "AHP" OR "Analytic Hierarchy Process" AND "Production Engineering" in three databases: Scopus, Web of Science and Scielo. All articles that used the AHP method to solve problems were counted for research done in April 2018, the oldest article being dated from April 1992 until the most recent one published, dated January 2018.
Then, the data were exported to the Microsoft Excel 2013 software and stratified into the following items: author, H index, co-authors, year, university, country, journal, CAPES qualis, impact factor, title, keywords, and digital base.
CAPES classifies the production from the quality of journals by the Qualis system, in a ranking by levels A1, A2, B1, B2, B3, B4, B5 and C and can be found on the Sucupira Platform (online). Its importance is determined through the note, which takes into account the quality and number of published works (BARATA, 2016). The area of Engineering III was chosen for the research. The impact factor of journals can be found in the publications' own databases, which take into account the number of citations received by articles published in the journal in the two years prior to the evaluation, divided by the number of articles published in the period.
The H index of researchers was also found in databases, which take into account the largest "h" number For the quantification of the indicators of years with the highest scientific production, the articles were ordered and counted per year of publication. The same method was used (ordering the appropriate variables for each index) to create the index of the most productive researchers, journals with more articles published on the topic, countries with more publications, more productive universities and keywords.
For the Cluster Word image, the keywords were placed on the website www.wordclouds.com, which generates clouds of words showing more prominently the words more frequently in the text.

V. RESULTS
The search terms resulted in 115 (one hundred and fifteen) articles from the Scopus database, of which 8 (eight) were withdrawn for not mentioning the method in the title of the paper or abstract. The Web of Science database returned 92 (ninety-two) articles, of which 21 (twenty-one) were excluded for not mentioning the method. And finally, the textual base Scielo, returned 0 (zero) articles. Unlike the previous databases that have articles in English, the Scielo database has many articles in Portuguese. An attempt was made to translate the term using "Multicriteria Decision Support Tools" OR "AHP" OR "Hierarchical Analysis Process" AND "Production Engineering" to search, but the result of the search again was 0 (zero ).    Table 3 shows the eight (n = 8) authors who had more than one publication in the research and its respective H index (according to the electronic database in which the file was found). Chan, FTS is the author with more publications, totaling 3 (three articles) and with an H index of 49 (forty-nine) points, being the author with the highest H index among the most productive authors.

International Journal of Advanced Engineering Research and Science (IJAERS)
[   IJPR is a well-established and highly successful journal reporting production and manufacturing research. It is published monthly and includes articles on manufacturing technology and the fundamental behavior of production resources, as well as the complex and interdisciplinary analysis and control problems that arise in combining these features in the design of production systems. The strategy of manufacturing, formulation and evaluation of policies and the contribution of technological innovation are the main concerns of the journal. Techniques developed in computational and mathematical sciences used in design, measurement or operation of production systems are also considered.
Although five publications were made in Brazil, no Brazilian journal appeared in this research (Table 5). Source: The author, 2018 Table 5 shows the ten countries that had the most publications on the AHP tool. The sum of the publications of these countries corresponds to 76.97% (137 articles) of the total of articles of the research.
China was the country with most publications, leading the ranking with 61 (sixty one) articles and a very significant difference of 44 articles for the second place, Turkey, with 17 (seventeen) publications. This is an interesting number for discussion, as it reiterates data from the 2018 edition of the journal Science & Engineering Indicators, which states that China surpassed the United States in the absolute number of articles produced, the country with the most published articles of the world.
Unlike the US that spends about $ 500 billion in research and technology but with near-frozen spending in recent years, China has been steadily increasing its investment in the area, currently spending $ 408 billion a year. In 2006, the country had approximately 190,000 published articles, and in 2016 reached the top position in the world leadership with more than 426,000, a substantial growth around 124%.  With a funding of $ 199.7 million, the university publishes more than 50 scientific journals and encourages teaching by giving full scholarships to students with the best marks in college entrance exams.
Istanbul Technical University is the oldest of the four, dating from 1773. With more than 20 institutes and research centers, the university was voted by QS World University Rankings as one of the 500 best university in the world and the best university in Istanbul in the field of Engineering and Technology.
Finally, Lappeenranta University of Technology began its activities in the year 1969. The university is known for awarding the Viipuri Prize, which boasts outstanding achievements in the field of strategic research.
Brazil had no prominent university in the subject, despite having 5 published articles, corresponding to 2.81% of the total articles found in the research. The articles totaled 1660 keywords, which represent on average 9.33 ± 3.12 keywords per article. The most prominent words were process (n = 89, corresponding to 5%), hierarchy (n = 69, corresponding to 4%), analytic (n = 60, corresponding to 4%), fuzzy (n = 49, corresponding to 3%) and production (n = 44, corresponding to 3%).
The high average for keywords is due to the fact that most definitions had multiple words. The similarity of the keywords can be explained by the directed nature of the research, which already had a defined thematic area to be explored.

VI. FINAL CONSIDERATIONS
The present article aimed to present a bibliometric study of publications between 1992 and 2018 on the method of decision support Analytic Hierarchy Process (AHP). The publications were classified by author, index H, co-authors, year, university, country, journal, qualis CAPES, impact factor, title, keywords, and digital base. Productivity indexes were then based on: year, author, journal, countries and keywords with the highest frequency.
The year 2013 was the year with the most publications, the most productive author was Chan, F.T.S with three publications on the subject and with the highest H index (49 points) among the most productive authors. The journal International Journal of Production Research appeared as the largest number of publications, China as the country with the most publications, four universities had the largest number of articles published, including Islamic Azad University, Istanbul Technical University, Lappeenranta University of Technology and University of Tehran and the keywords that appeared most frequently in the text were process, hierarchy, analytic, fuzzy and production.
This work showed the importance of the AHP method with regard to decision making, specifically focused on the area of Production Engineering. More than 178 researchers relied on their accuracy to prioritize and solve problems, showing that even after 46 years the method still shows relevance. this fact, the work had a more directed nature and found similar examples.
It is suggested for future research that the thematic area should be broader, not limited to Production Engineering but to other Engineering, or even open research. It would be interesting also to make a bibliometric comparison between other methods of support to decision making, coming from a distinction between date of creation or even schools of thought.