Introduction
Semantic search behavior refers to a set of cognitive and practical processes and actions undertaken by users during information searching, in which the individual goes beyond mere word matching to understand and retrieve information based on meaning, concept, and conceptual relationships among terms. The purpose of this study was to analyze the semantic information search behavior of faculty members at Shahid Modarres Hospital in PubMed regarding COVID-19, based on Ingwersen's model.
Research Questions
1. How does the semantic search behavior for COVID-19 information in PubMed by faculty members of Shahid Modarres Hospital occur based on the 'information retrieval systems' component in Ingwersen's cognitive model?
2. How does the semantic search behavior for COVID-19 information in PubMed by faculty members of Shahid Modarres Hospital occur based on the 'identification of information content' component in Ingwersen's cognitive model?
3. How does the semantic search behavior for COVID-19 information in PubMed by faculty members of Shahid Modarres Hospital occur based on the 'user interface' component in Ingwersen's cognitive model?
4. How does the semantic search behavior for COVID-19 information in PubMed by faculty members of Shahid Modarres Hospital occur based on the 'cognitive space' component in Ingwersen's cognitive model?
5. How does the semantic search behavior for COVID-19 information in PubMed by faculty members of Shahid Modarres Hospital occur based on the 'social/organizational environment' component in Ingwersen's cognitive model?
Literature Review
Previous studies indicate that information search behavior and semantic search are influenced by individual, systemic, and contextual factors. Shohrati (2022) reported that physicians and nurses made the most use of the PubMed database for searching COVID-19-related information, and their primary objectives included understanding the virus, fulfilling information needs, and sharing information. Khoshian et al. (2024) showed that although semantic search has moderate performance in conceptual connectivity and retrieval of multi-word concepts, it contributes to improving information retrieval. Khosravi et al. (2015) found no significant difference between PubMed's suggested keywords and MeSH terms in terms of result relevance, although both contribute to the development of search vocabulary. Lazarus et al. (2020) demonstrated that combined strategies based on free text and MeSH yield the highest sensitivity in retrieving COVID-19 articles. Furthermore, studies by Das and Saha (2019), Nemeshaev et al. (2021), and Shiri & Revie (2006) have emphasized the role of semantic search, ontology, and user experience in enhancing retrieval accuracy and improving query expansion.


Methodology
The present study is applied in terms of purpose and was conducted with a quantitative approach using a descriptive-analytical survey method. The research population consisted of 64 faculty members of Shahid Modarres Hospital. Due to the limited size of the population, no sampling was performed, and 54 questionnaires were collected (response rate of 84%). The data collection instrument was a researcher-developed questionnaire designed based on Ingwersen's cognitive model and the semantic search capabilities of PubMed. The content validity of the instrument was reviewed and confirmed by experts in information retrieval, semantic search, and clinical librarianship, and its reliability was obtained with a Cronbach's alpha coefficient of 0.80. Data were collected both in-person and online and were analyzed using SPSS software version 25.
Results
Faculty members actively use PubMed's semantic search features, including simple search, auto-suggestions, term refinement, combined strategies, MeSH (especially Major Topics), synonyms, and suggested keywords. While they are generally satisfied with PubMed's layout, use of frameworks like PICO and advanced indexing tools (e.g., Show Index List) is limited. Abstracts are the primary tool for assessing relevance, while personalized features like My NCBI are underutilized. Users rely on full texts for conceptual understanding, with less use of citation tools. Boolean operators and user-friendliness are rated favorably. Search behavior is influenced by scientific, research, and clinical needs, prior experience, database awareness, and the need for authoritative information. PubMed helps reduce uncertainty and supports decision-making. Organizational factors such as infrastructure, library services, and professional priorities also shape search behavior. Significant differences in semantic search behavior were found based on academic rank, years of service, and expertise, but not by gender.
Discussion
The research findings indicate that faculty members largely utilize PubMed's semantic capacities for search development. This result is consistent with the study by Khoshian et al. (2024), which confirmed the role of semantic search in accessing more relevant scientific information. Additionally, the use of MeSH and PubMed's suggested keywords aligns with the findings of Khosravi et al. (2015), who considered these tools effective in expanding users' vocabulary range.
Despite the appropriate use of some semantic capabilities, the results showed that more advanced tools such as MeSH Subheadings, Show Index List, and the PICO framework have not yet been fully institutionalized in users' search behavior. This finding is consistent with the results of studies by Eriksen & Frandsen (2018) and Fineout-Overholt et al. (2024) regarding the limited use of structured search frameworks by users. The limited use of personalized features such as My NCBI indicates that a portion of this database's individual knowledge management capacities has not yet been fully utilized.
According to Ingwersen's cognitive model, the results confirm that semantic search behavior is not merely a function of system features, but rather the interaction among users' cognitive characteristics, professional goals, and organizational environment also plays a role in its formation. Search experience, information needs, and users' scientific and clinical objectives determine the path of selecting search strategies.


Conclusion
The results of this study showed that the semantic search behavior of faculty members in PubMed is the result of the interaction among system capabilities, users' cognitive factors, and socio-organizational conditions. Although users make good use of basic features and some of PubMed's semantic tools, there is a need for further training in utilizing advanced tools such as MeSH Subheadings, PICO, indexing features, and personalized capabilities. Accordingly, designing targeted educational programs to enhance semantic search skills, developing visual guides for the use of search frameworks, and strengthening the role of clinical librarians can lead to improved information retrieval quality and more effective support for scientific and clinical decision-making.