Introduction

Regarding the increase in digital images and easy access to digital cameras, image processing and retrieval have become an important research field. Image retrieval is one of the important subfields of information retrieval that usually uses different techniques and models than text retrieval. The main expectation of users of information retrieval systems is to find relevant resources among thousands of resources available in these systems. Creating a scientific map of articles in the field of image retrieval using ontology can provide awareness of the status of published research; it can also help to show thematic relationships, identify influential topics, mature, emerging, and undeveloped topics, thematic gaps, and create appropriate scientific policies in this field.

Literature Review

A small number of studies have been conducted in the field of image retrieval using ontology, none of which have examined the hierarchical diagram. In the field of scientometric research, such as Daniali, Naghshineh, and Fadaei (2017); Daniali and Naghshineh (2018); Azimi and Jozi (2014); Ghanbari et al. (2014); Liu et al. (2021); Jo (2024); Khan et al. (2024). Scientometric studies often focus on assessing publication patterns, citation networks, co-authorship, and research productivity. These studies help researchers understand the structure and evolution of the scientific literature in a given field.

Methodology

In terms of type, the present study is in the category of applied research in which scientometric techniques and social network analysis have been used. The research community consists of those studies in the field of information retrieval with ontology that have been indexed in the Scopus database from the beginning to the end of 2024.
Based on the formula in the search, 716 indexed articles were found in Scopus in the desired field. To be more precise, the statistical population of this study consists of all 716 published articles in the field of image retrieval using ontology.
After retrieving relevant records and integrating data, based on the research objectives and questions, data analysis was carried out using BibExcel, Gephi, Excel, and SPSS, and the maps were created by VOSviewer software. In order to draw thematic maps and analyze them correctly, keywords were controlled and standardized by creating a Thesaurus in the software. In such a way that identical and similar keywords and plural and singular forms were merged and non-specialized and searched keywords were removed. In order to classify words in published documents based on semantic similarity using algorithms such as Euclidean distance and..., hierarchical clustering is usually used. Hierarchical clustering was performed using SPSS software. In order to implement and achieve the analysis, requirements such as a co-occurrence matrix must first be prepared, and then the co-occurrence matrix must be converted into a correlation matrix. The statistical population of the present study was the entire population. In order to perform a more accurate synonym analysis and final synonym analysis, the matrix was called through SPSS software, and the regular matrix was converted into a correlation matrix by SPSS software. The correlation matrix that was based on the obtained cognates frequency matrix, clusters and hierarchical were drawn using hierarchical clustering using the Ward method and squared Euclidean distance. Among the 716 retrieved articles, keywords with a frequency of 11 and more were selected for the research to prepare the matrix, and finally a square matrix of 142 by 142 was formed for the research.
The diagonal cells of the matrices were considered zero and then these ordinary matrices were converted into a correlation matrix. Finally, the clustering of concepts was drawn based on statistical software (SPSS version 26). Finally, to draw a scatter diagram and identify the development and maturity status of the topics, the frequency matrix of each cluster was drawn separately, then their correlation matrix was drawn, and with the help of the strategic diagram, using the density and centrality of each cluster, their coherence and maturity were calculated. In the next step, a strategic diagram of thematic clusters was drawn; the strategic diagram describes the internal relationship and correlation between the different thematic clusters. Excel, Gephi, Babel Excel, and SPSS software were used to analyze the data, and Word Viewer, Excel, and SPSS software were used to draw the diagram.

Results

The results showed that among the keywords, the keyword image retrieval is in the first place with a frequency of 572. Among them, the keywords ontology and semantics are in the second and third places, respectively. Also, the analysis of the findings related to the synonym of image retrieval using ontology took the form of six thematic areas. Infrastructures and fundamental technologies of image retrieval, techniques of concepts and image analysis, intelligent machines and applications in image retrieval, web concepts and search based on ontology, managerial and human aspects in information retrieval, quality improvement and query processing in image retrieval.
The results obtained from the hierarchical diagram formed four topic clusters. Also, the findings from the strategic map of image retrieval topics using ontology indicate that cluster 1 was placed in the first part due to its high centrality and density. These clusters are higher in centrality and density. Clusters 2 and 4 are placed in the second part. The clusters that are located in the second part of the strategic picture are regional clusters. But they are developed. Cluster 3 is located in the fourth part. The clusters in the fourth part are the main ones, but are undeveloped and immature.

Discussion

With the help of scientometrics, a macro picture of the state of research and how different domains are related, can be presented. Co-occurrence analysis led to the formation of six clusters. Fundamental infrastructures and technologies of image retrieval, semantic techniques and image content analysis, machine learning and intelligent applications in image retrieval, semantic web and ontology-based search, managerial and human aspects of information retrieval, quality improvement and query processing in image retrieval. Among the existing concepts, some of them have received the highest number of citations: such as image retrieval, machine learning, computer vision, deep learning, database systems, digital libraries, Internet, WordNet, artificial intelligence, and information retrieval. A hierarchical diagram of four thematic clusters was formed: semantic-based image retrieval, intelligent image retrieval with learning algorithms, semantic image retrieval with annotations, and intelligent image retrieval.

Conclusion

Studies in the field of image retrieval using ontology as a useful tool for effective retrieval can play an effective role. Most studies are in the field of ontology and ontology construction, but no specific research has specifically addressed this area. In this study, the emphasis is on image retrieval, but the high importance of this area, the recognition of the components of this area, and the impact of ontology on semantic retrieval require researchers to focus on these issues, and the results of this study indicate that very little attention has been paid to this area, especially in Iran. Scientific maps are a suitable method for displaying the increasing growth of scientific activities and organizing the intellectual and scientific structure that constitutes a thematic domain. Researchers, science policymakers, and other interested parties can advance their own goals and advance with greater awareness in this field by being aware of this structure.