Delivering services in digital libraries must align with users’ preferences, needs, and feedback. In this context, service personalization plays a pivotal role in enhancing user satisfaction and maximizing the utilization of digital library capacities. This study aims to systematically review the existing research on service personalization in digital libraries to identify key developments, research gaps, and future directions. Employing a qualitative approach, the study follows the systematic review framework proposed by Kitchenham and Charters. A total of 67 sources —including journal articles, conference papers, and master’s and doctoral theses— were identified and analyzed. Findings reveal that the research landscape can be categorized into three main dimensions and nineteen subcategories: 1) user-centric services (including user profiles, user modeling and privacy, user behavior, collaborative environments, personal information spaces, contextual interoperability, user context awareness, user services, information security, cognitive styles, access methods, and “My Library” features), 2) types of personalization (covering personalization processes, indicators and methods, personalized search and results, recommender systems, and filtering), and 3) applied techniques and technologies (such as data mining, intelligent technologies, big data, and cloud computing). Among these, recommender systems received the most attention. The evolution of research in this domain reflects a transition from foundational digital library infrastructure to advanced intelligent personalization, encompassing AI-driven behavior prediction, interest recognition, automated analysis, and customized service delivery. Despite notable progress, the study highlights the need for innovative and diverse research to address emerging challenges and technological shifts.
Conversational Question Answering (CQA) systems have evolved significantly with the advent of Large Language Models (LLMs). However, these advancements have predominantly favored high-resource languages, often overlooking low-resource ones. This paper introduces a novel LLM-enhanced framework specifically designed to bridge this linguistic gap. The proposed architecture comprises six components: "Input Processing" for language-specific handling, an "Adaptive LLM Core," "Knowledge Enhancement" for cross-lingual mapping, "Context Management" for efficient conversation navigation, "Response Generation" incorporating cultural adaptation, and "Human Feedback" for continuous improvement. Unlike existing approaches, this framework integrates cultural and linguistic considerations throughout the entire processing pipeline. To validate the framework, a qualitative evaluation was conducted using a focus group consisting of five Natural Language Processing (NLP) experts. Expert evaluation results confirmed the proposed framework's effectiveness in addressing fundamental challenges of low-resource languages, including data scarcity, morphological complexities, and cultural nuances. Experts particularly highlighted the framework's innovative approach to "integrated cultural processing," "resource efficiency" via optimized context management, and its "modular and scalable architecture" as key achievements. This research demonstrates that integrating human feedback and cultural adaptation within an efficient architecture offers a practical solution for developing Conversational Question Answering systems in low-resource languages.
This study aims to analyze the conceptual structure of the Semantic Web domain within Knowledge and Information Science (KIS) using data from the Web of Science (WoS) database spanning from 1995 to 2024. The study was conducted using quantitative content analysis and network analysis methods. The research population comprised 1,761 articles. Data analysis was performed using VOSviewer, UCINET, and BibExcel software. Following keyword standardization, a co-occurrence matrix was constructed, and concepts were categorized into ten thematic clusters using the K-means clustering algorithm. Subsequently, a strategic diagram was plotted based on centrality and density indices. The Semantic Web domain has evolved into a dynamic, multi-layered structure over the last three decades. The plotted strategic diagram offers a clear, evidence-based image of conceptual maturity, internal cohesion, and the epistemological status of research clusters. This conceptual map can serve as an analytical and policy tool for scientific decision-makers and researchers to identify knowledge gaps, determine research priorities, and guide future research paths more accurately and efficiently. The analysis identified ten clusters: (1) Ontology and Semantic Retrieval, (2) Semantic Web and Knowledge Management, (3) Linked Data and Information Organization, (4) Thesauri and Knowledge Organization Systems (KOS), (5) Digital Cultural Heritage and Knowledge Organization, (6) Metadata and Knowledge Representation, (7) Semantic Web Standards and Languages, (8) Semantic Information Retrieval in the Web Environment, (9) Open Science and Knowledge Engineering, and (10) Open Data Integration. The clusters of Thesauri and KOS, Digital Cultural Heritage, Metadata and Knowledge Representation, Open Science, and Open Data Integration were identified as emerging fields. The average growth rate of articles during the studied period was 57%. The concepts of Semantic Web, Ontology, and Linked Data comprised of the three main pillars and the central core of this research domain, with the highest frequencies. Notably, the significant presence of keywords from applied fields shows the penetration and influence of the Semantic Web in other areas. Analysis of the strategic diagram indicates that Linked Data, Information Organization, and Semantic Information Retrieval, positioned in the strategic quadrant, constitute the mature core of Semantic Web research. The findings reveal a shift in this domain from an exclusive focus on technical standards and languages—which have now achieved a foundational and established status—towards operational implementation and intelligent knowledge management. This structural evolution, consistent with trends observed in prior literature, signifies a paradigm shift in the Semantic Web from an infrastructure development phase to a practical application within information systems.
In recent years, university evaluation and ranking have become primary concerns for higher education institutions and policymakers. Research criteria play a vital role in determining the academic standing of universities. However, one of the fundamental challenges in this area is determining the relative importance of research criteria and accurately modeling their relationship with university rankings. Many studies have employed classical statistical methods or multi-criteria decision-making techniques but often lack a precise integration between objective criteria weighting and analyzing their impact on ranking levels. This research utilizes a hybrid approach and analyzes real-world data on the research performance of Iranian universities to address existing gaps in the effective integration of weighting and impact analysis of criteria. In the first step, the CRITIC method, as an objective and data-driven technique, is used to determine criterion weights. This method simultaneously considers the dispersion and correlation of criteria to establish their relative importance without subjective judgment. Subsequently, the obtained weights are incorporated into an Ordinal Logistic Regression model to examine the impact of each criterion on the probability of universities being placed in different ranking levels. The main innovation of this study lies in the structured combination of these two complementary methods and the focus on analyzing the coefficients of the ordinal logistic regression model. This approach, unlike some superficial methods, provides a more accurate picture of the role of each criterion. The results indicate that, contrary to expectations, some criteria do not have a significant impact on improving university rankings, while specific, others have a considerable effect. The research findings can be utilized by higher education managers and policymakers in designing research policies, optimizing research budgets, and planning for enhancing the academic standing of universities. This framework is also generalizable to other domestic and international ranking systems.
The aim of this study is to design a contingency and conceptual model of content marketing for Iranian reading applications utilizing two theoretical lenses: customer value theory and institutional theory.This study employed a qualitative approach using thematic analysis. Data were collected through semi-structured interviews with 10 content marketing experts familiar with reading applications. Purposive snowball sampling continued until theoretical saturation was achieved. Data analysis was performed using MAXQDA software in six steps, and coding reliability was confirmed with a Holsti coefficient of 0.82.The final model consists of three main parts: 1) Dimensions of value creation for the user, including functional values (ease of access), emotional values (enjoyment of reading), social values (sharing), cognitive values (learning), and conditional values (using dead time); 2) Institutional constraints of Iran's environment at three levels: coercive pressures (filtering, sanctions), normative pressures (self-censorship, publisher monopoly), and mimetic pressures (copying competitors); 3) Operational elements of strategy including targeting, multi-format content production (podcasts, videos, text), targeted distribution (Instagram, the app itself), user engagement, and performance measurement using indicators such as engagement rate and dwell time.Successful content marketing in Iranian reading applications requires a balance between creating value for the user and adapting to institutional constraints. By identifying moderating variables (app characteristics and user characteristics), the proposed model goes beyond common linear models and provides a flexible framework for developing a strategy tailored to the conditions of each application.
