ارزیابی کیفیت پایگاههای داده علمی با استفاده از الگوریتمهای هوش مصنوعی: مطالعه موردی پایگاههای فارسی و بین المللی

of intelligent information infrastructures and the use of new technologies in scientific Scientific databases, as one of the most important components of the information infrastructure of the scientific and research system, play a decisive role in the production, organization, retrieval, and evaluation of knowledge. These databases are the basis for conducting scientific research, scientometric analyses, evaluating the performance of researchers and scientific institutions, and formulating macro-scientific policies at the national and international levels. The quality of data and metadata recorded in these databases has a direct impact on the validity of research results, the accuracy of quantitative scientometric analyses, and the effectiveness of evidence-based decision-making. Given the increasing growth of scientific production, the increasing diversity of information sources, and the expansion of Persian and international databases, the need for systematic, accurate, objective, and scalable approaches to assess the quality of these databases is increasingly felt. Traditional methods for assessing the quality of databases, which are mainly based on human judgments and limited sample surveys, do not have the necessary efficiency and accuracy due to the large volume of data and the complexity of the information structure.

Objective: The main objective of this research is to develop a new and scientific framework based on artificial intelligence algorithms for assessing the quality of scientific databases. This study attempts to identify and analyze the strengths and weaknesses of Persian databases in comparison with reputable international databases, while providing a clear picture of the current situation, and to provide scientific solutions for improving the quality of information services of these databases and to provide a reliable framework for use by researchers, database managers, and policymakers in the field of scientometrics.



Method: The present study is applied in terms of its purpose and descriptive-analytical in terms of its nature and method. The research community includes Persian and international scientific databases active in the field of indexing and dissemination of scientific products. In this study, Persian databases including the Scientific Information Database of the Academic Jihad (SID), Mag Iran, and Normags, and international databases including Scopus and Web of Science were selected as the samples under study. Research data were collected by extracting bibliographic records and metadata related to scientific articles available in this database. In order to assess the quality of the databases, a set of key quality indicators were determined, including accuracy, data, information comprehensiveness, metadata consistency and standardization, data up-to-dateness, information retrieval capability, and duplicate data rate. For data analysis and objective evaluation of the indicators, artificial intelligence and machine learning algorithms were used to analyze metadata errors, identify duplicate records, evaluate the similarity of descriptive information, and measure the efficiency of information retrieval. The use of these approaches enabled the processing of large volumes of data, reduced the interference of human judgment, and increased the accuracy and scalability of the quality assessment process.



Findings: The findings of the study showed that international databases perform better in most quality indicators. Specifically, the average data accuracy in Scopus was estimated to be 96 percent and in Web of Science to be about 95 percent, while these indicators were on average 85 percent in Persian databases. In terms of metadata integrity, international databases scored 94 percent, while this score was reported to be 80 percent for Persian databases. Also, the rate of duplicate data in Persian databases was estimated to be 5 percent on average, while this rate was less than 2 percent in international databases. In the information up-to-dateness index, international databases with an average time delay of less than three months performed better than Persian databases with a 5-month delay.

Conclusion: The present study shows that the use of artificial intelligence algorithms can be an effective tool for objective and scalable assessment of the quality of scientific databases. Although Persian databases play an important role in accessing local scientific productions, they need to improve the metadata structure, reduce information errors, accelerate the update process, and widely use smart technologies to achieve international standards. The framework presented in this study can be used as a scientific basis for the optimal selection of databases by researchers and also to improve scientific information policies. The application of this framework can lead to improved metadata quality, reduced errors and duplicate data, increased data retrieval capabilities, and ultimately the credibility and efficiency of scientific databases, especially Persian databases. This study emphasizes the need for targeted investment in the development knowledge management.