Abstract:

This study is aimed to predict the expected publication performance in a decade, while evaluating the factors that affect the change in the number of publications in Turkey, the effect of these factors, and whether citation indexes will be used as a criterion to monitor scientific progress. Scientific publication performance is influenced by a country's population, universities, and academic staff, the number of students specialising in their field, and the proportion of the gross national product allocated to research and development. In determining the scientific standing of nations, comparing the scientific credentials of nations or universities, and assessing the academic performance of scientists, the following criteria, emphasising "international publication activities," are widely accepted:
i) the number of articles published in international scientific journals;
ii) the number of scientific indexes of journal articles; and
iii) the number of citations to articles.

Significant studies have been conducted in recent years to evaluate the international scientific publications of Turkish universities. Two reports published by the Turkish Academy of Sciences are included.

A generalized regression neural network (GRNN) model was employed for the forecasting in this research. Artificial neural networks (ANNs), a sort of information processing system modeled after the mathematical abstractions underlying human neurons, are capable of inferring correlations across different data sets based on existing patterns.This technique is used to predict the future value of numerous variables simultaneously and simulate erroneous interactions involving large amounts of data. On the other hand, bibliometric analysis on the CiteSpace platform quantitatively highlights the major aspects of literary works and their uses. The study intends to give an in-depth literary evaluation using bibliometric analysis and to indicate what types of topics are prominent in Turkey. Unlike other review publications, this study analyses the research area using a text mining technology called CiteSpace tool. Data on the total number of publications addressed by Turkey between 1984 and 2015 is used to train, validate, and test the network. The input data was formed using citation indexes prepared by the Scientific Information Institute. The National Academic Network and Information Center used the International Citation Index (ULAKBIM) SC to examine the number of publications Turkey addressed in the index from 1981 to 2015. SSCI and AHCI broadcasts were located in the Web of Science database on the web page. Then, the number of publications, number of citations, and impact values were found in the evaluation reports of the International Scientific Publication Incentive Program. The most crucial part of the process, data collection, was dependent on various types of documentation. Among the obtained publications are articles, letters to the editor, notes, book reviews, abstracts, reviews, and revisions. The scores of the ANN model are presented in the results section as mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and R-square value, respectively. The bibliometric analysis findings are also presented graphically. In the concluding section, we consider future directions for the developed model by highlighting emerging research problems and emerging trends in the field.

Keywords:

Scientometric Analysis, Artificial Neural Networks, Educational Big Data, Artificial Intelligence