BookNexus

Journal of Advancement in Software Engineering and Testing Page No. 19 http://www.hbrppublication.com 2026: 9 (1), 19-24 e-ISSN: 2584-2226 Volume 09 Issue 01 Jan-Apr, 2026 *Corresponding Author: Sudhan Sanjay V P, Student, Department of Computer Science & Engineering, Sri Shakthi Institute of Engineering and Technology, Coimbatore, Tamil Nadu, India BookNexus 1Sanjay Nithin S, 2Sudhan Sanjay V P*, 3Vignesh K, 4Ms. Sini Prabhakar 1-3 Student, 4Assistant Professor, Department of Computer Science & Engineering, Sri Shakthi Institute of Engineering and Technology, Coimbatore, Tamil Nadu, India ABSTRACT Efficient information retrieval and personalized learning have become essential aspects of modern digital education systems. BookNexus: AI-Powered Library is designed to enhance the process of book discovery and knowledge access through advanced artificial intelligence techniques. The system employs semantic search combined with Retrieval-Augmented Generation (RAG) to deliver contextually relevant results, improving search precision beyond traditional keyword-based methods. It further personalizes user experiences by analyzing reading behavior and interest patterns to recommend suitable materials. To maintain credibility, integrated modules detect spam and fake reviews, ensuring authenticity in usergenerated content. Face recognition is implemented for secure and seamless access, while offline voice search enables accessibility across varied environments. Interactive dashboards provide dynamic insights for both users and administrators. Developed using Django, ARC Face, and contemporary web technologies, BookNexus demonstrates the potential of AI-driven digital libraries to transform information management and personalized learning in the modern era. Keywords:- Digital library, Semantic search, RetrievalAugmented Generation, Personalized recommendation, Artificial intelligence, Learning analytics Journal of Advancement in Software Engineering and Testing Page No. 20 http://www.hbrppublication.com 2026: 9 (1), 19-24 1. INTRODUCTION In recent years, the rapid advancement of artificial intelligence (AI) and data-driven technologies has transformed how digital content is accessed, managed, and personalized. Traditional library systems often struggle to meet the growing demands of intelligent search, reliable reviews, and adaptive user experiences. BookNexus: AI-Powered Library addresses these challenges by integrating advanced AI techniques—including Machine Learning (ML), Natural Language Processing (NLP), and Retrieval-Augmented Generation (RAG)—to modernize the way users interact with digital libraries. The exponential growth of digital knowledge repositories and user-generated data has made conventional keyword-based search methods insufficient for providing relevant and context-aware results. To overcome these limitations, BookNexus employs semantic search and vector-based retrieval, enabling the system to understand the meaning and context of user queries rather than relying solely on textual matches. This approach significantly enhances book discovery and ensures precise information retrieval, particularly for academic and researchoriented users. Furthermore, personalized recommendations form a core component of the system. By analyzing user history, behavioral patterns, and preferences, BookNexus provides intelligent reading suggestions that evolve dynamically over time. The inclusion of spam and fake review detection models strengthens the trustworthiness of content and prevents manipulation in book ratings or feedback. In addition, the integration of ARC Face recognition ensures secure user authentication, while offline voice search extends accessibility to environments with limited internet connectivity. From an administrative perspective, interactive dashboards provide realtime analytics on user activity, content popularity, and library usage trends. These insights support data-driven decision-making for improving library operations and enhancing user engagement. The system’s modular design—comprising input handling, semantic retrieval, personalized recommendation, spam detection, analytics, and response modules—ensures scalability, reliability, and ease of integration with existing digital infrastructures. Overall, this project contributes to the growing field of intelligent information management by presenting a comprehensive framework that unites AI, ML, and NLP to deliver context-aware, personalized, and secure library experiences. By reimagining how users search, discover, and engage with digital content, BookNexus establishes a new paradigm for AI-powered learning and knowledge ecosystems. 2. EXISTING SYSTEM Traditional digital library systems mainly serve as static repositories that store and retrieve digital content through basic keyword-based searches. These conventional approaches often emphasize database management and manual cataloging rather than intelligent knowledge discovery or contextual understanding. As a result, users frequently face challenges in locating relevant materials, since the system fails to interpret the semantic meaning of queries or adapt to user intent. Existing library management systems lack mechanisms for personalized recommendation or predictive analytics. Recommendation features, if available, are generally limited to simple metrics such as item popularity or manual tagging rather than analyzing user preferences, reading habits, or contextual relationships. Consequently, users receive repetitive or irrelevant suggestions, reducing engagement and efficiency. Journal of Advancement in Software Engineering and Testing Page No. 21 http://www.hbrppublication.com 2026: 9 (1), 19-24 Furthermore, most platforms depend on user-generated reviews without verification, making them vulnerable to spam or fake feedback. This undermines the credibility of book ratings and recommendations. In addition, security within these systems largely relies on traditional password-based authentication, which poses privacy risks and fails to provide a seamless access experience. Administrative functions in current systems are mostly restricted to record keeping and report generation. They lack real-time dashboards or analytical insights into user behavior, book trends, or operational efficiency. Therefore, existing systems remain reactive rather than intelligent or adaptive. To overcome these limitations, an advanced, AI-driven framework like BookNexus is required to enable semantic search, secure authentication, spam detection, and personalized recommendations—ensuring a smarter and more reliable digital library experience. 3. PROPOSED SYSTEM The proposed system, BookNexus: AIPowered Library, leverages artificial intelligence, machine learning, and natural language processing to transform the digital library experience into a more intelligent, secure, and personalized ecosystem. It addresses the shortcomings of traditional systems by integrating semantic search, Retrieval-Augmented Generation (RAG), face recognition, and spam detection into a unified platform. The system operates through multiple modules: input processing, semantic retrieval, recommendation, spam and review detection, analytics, and real-time response generation. User queries— whether entered via text or voice—are semantically analyzed using RAG and sentence-transformer embeddings to provide contextually relevant search results. The system’s recommendation engine personalizes suggestions based on user reading patterns, history, and preferences. Spam and fake review detection models ensure authenticity by filtering unreliable content through trained classifiers. Secure access is achieved using ARC Face recognition, while offline voice search enhances accessibility across varied environments. Additionally, interactive dashboards display analytical insights, including book popularity, user engagement, and system performance. Key Functionalities:  Semantic Search & Retrieval: Uses RAG and sentence embeddings for context-aware, precise search results.  Personalized Recommendation: Analyzes user behavior and preferences to suggest relevant books dynamically.  Spam & Review Detection: Identifies and filters fake or malicious reviews using trained ML models.  Face Recognition Login: Ensures secure, biometric-based authentication for users.  Analytics Dashboard: Provides realtime visualizations for administrators and users.  Offline Voice Search: Enables voicebased interaction without constant internet access. Advantages of the Proposed System: ● Intelligent and Context-Aware: Delivers semantically accurate results using AIdriven techniques. ● Secure and Reliable: Incorporates face recognition for user verification and spam filtering for authenticity. ● Personalized Experience: Adapts dynamically to user interests for improved engagement. ● Data-Driven Insights: Real-time dashboards assist administrators in informed decision-making. ● Scalable and Adaptive: Modular design allows easy integration of new AI models and features. ● Offline Voice Search: Enables voice- Journal of Advancement in Software Engineering and Testing Page No. 22 http://www.hbrppublication.com 2026: 9 (1), 19-24 based interaction without constant internet access. 4. SYSTEM ARCHITECTURE The proposed system follows a modular architecture designed to enhance digital library operations through artificial intelligence and machine learning. It consists of six primary components: Input Processing, Data Preprocessing, Semantic Retrieval, Recommendation Engine, Spam & Review Detection, and Analytics & Response Module.  Input Processing – Accepts user input through text or offline voice queries and converts them into structured data for semantic analysis.  Data Preprocessing – Cleans and normalizes text by removing special characters, stop words, and irrelevant symbols. Tokenization and lemmatization are applied before encoding data using Sentence Transformers to generate vector embeddings.  Semantic Retrieval – Implements Retrieval-Augmented Generation (RAG) to perform intelligent, context-aware searches across the library database, retrieving results based on meaning rather than keyword similarity.  Recommendation Engine – Analyzes user history, reading patterns, and preferences to provide personalized book suggestions dynamically.  Spam & Review Detection – Utilizes trained ML classifiers to identify fake or misleading reviews, ensuring authenticity and credibility of user feedback.  Analytics & Response Module – Delivers real-time dashboards for administrators and users, visualizing trends in book popularity, user engagement, and system performance to support data-driven decision-making. 5. PREPROCESSING To ensure accurate and efficient performance of the BookNexus: AIPowered Library system, several preprocessing steps are applied to prepare textual and user-generated data for machine learning and semantic analysis.  Data Cleaning – Removes unnecessary elements such as HTML tags, special symbols, and duplicate entries from book descriptions, user queries, and reviews to maintain data consistency and quality.  Text Normalization – Converts all text to lowercase and standardizes formatting, improving uniformity across diverse data sources.  Tokenization and Lemmatization – Breaks sentences into individual tokens and reduces words to their root forms to enhance the semantic understanding of text during model training.  Encoding – Transforms processed text into numerical vector embeddings using Sentence Transformers (MiniLM-L6-v2), enabling semantic similarity comparison for search and recommendation.  Feature Extraction and Selection – Derives essential features such as sentiment polarity, spam probability, and contextual embeddings while discarding irrelevant or redundant attributes to improve model efficiency.  Data Segmentation – Organizes datasets for training, validation, and testing phases to ensure optimal evaluation of recommendation accuracy and spam detection performance. 6. FEATURE EXTRACTION In the BookNexus: AI-Powered Library system, feature extraction is essential for enabling semantic search, recommendation, and spam detection. Textual and behavioral data are processed to identify meaningful attributes that improve model accuracy and efficiency. ● Text Embeddings: Book titles, descriptions, and user queries are encoded into vector embeddings using Sentence Transformers (MiniLM-L6-v2), allowing context-based similarity comparisons. ● Sentiment and Spam Indicators: Journal of Advancement in Software Engineering and Testing Page No. 23 http://www.hbrppublication.com 2026: 9 (1), 19-24 User reviews are analyzed to determine sentiment polarity (positive, neutral, or negative), while spam-related features such as repeated patterns and linguistic irregularities help identify fake content. ● User Interaction Patterns: Reading frequency, search behavior, and feedback activity are captured to support personalized recommendations. By selecting only the most relevant textual and behavioral features, the system ensures faster computation and improved accuracy in AI-based processing. 7. CLASSIFICATION The BookNexus system employs machine learning models to classify and refine search results, recommendations, and reviews. ● Spam Review Detection: Reviews are classified as genuine or spam using trained models such as Logistic Regression, Decision Tree, and Random Forest, with Random Forest yielding the best accuracy. ● Sentiment Classification: Reviews are categorized as positive, neutral, or negative to enhance recommendation relevance. ● Recommendation Ranking: Books are ranked as highly, moderately, or less relevant based on user history and semantic similarity. This multi-level classification ensures precise search outcomes, trustworthy feedback, and adaptive personalization. 8. RESULT The performance of BookNexus: AIPowered Library was evaluated using preprocessed datasets of books, user queries, and reviews, with a 70:30 training-testing split for machine learning models. Key modules—including spam detection, sentiment analysis, and recommendation ranking—were assessed using standard metrics such as accuracy, precision, recall, and F1-score. Among the models implemented, Random Forest achieved the highest performance across multiple tasks, providing the best balance between precision and recall. Logistic Regression, Decision Tree, and Support Vector Machine models were also tested, showing slightly lower accuracy in identifying spam reviews and ranking relevant recommendations. Feature importance analysis indicated that semantic embeddings, user interaction patterns, and review sentiment were the most influential factors in improving recommendation relevance and content authenticity. The results demonstrate that BookNexus can effectively provide context-aware search results, reliable review classification, and personalized book recommendations. Additionally, the system’s modular architecture allows easy integration of new datasets, adaptation to emerging AI models, and scalable handling of growing user bases. These capabilities ensure that both users and administrators can benefit from a dynamic, responsive, and trustworthy library platform. Applications ● Personalized book discovery for users based on interests and reading history. ● Detection and filtering of spam or fake reviews to maintain content credibility. ● Real-time analytics and dashboards for library administrators. ● Voice-based search and offline accessibility for improved usability. ● Data-driven decision-making for library management and content planning. ● Enhanced engagement tracking to inform content updates and library services. 9. CONCLUSION The BookNexus: AI-Powered Library system demonstrates the effectiveness of AI and machine learning in enhancing digital library experiences. By leveraging Journal of Advancement in Software Engineering and Testing Page No. 24 http://www.hbrppublication.com 2026: 9 (1), 19-24 semantic search, personalized recommendations, sentiment analysis, and spam detection, the system delivers context-aware results, reliable content, and a secure user experience. Evaluation metrics confirm that Random Forest and other machine learning models provide accurate and consistent classification for spam detection and recommendation relevance, ensuring user trust and engagement. This system empowers administrators with actionable insights through interactive dashboards and analytics while offering users a personalized and accessible library environment. Its scalable design allows future integration of additional AI features, expanded datasets, and crossplatform accessibility, ensuring long-term usability and adaptability. Future Scopes: ● Integration of advanced behavioral analytics to further refine personalized recommendations. ● Expansion of offline voice search and multi-lingual support for broader accessibility. ● Incorporation of adaptive learning pathways to suggest content based on user proficiency and interests. ● Continuous improvement of spam and fake review detection using larger, evolving datasets. ● Implementation of recommendation explainability features to help users understand why content is suggested. ● Integration with institutional library systems for automated catalog updates and notifications. Overall, BookNexus establishes a scalable and intelligent framework for modern digital libraries, combining AI-driven insights with practical usability. REFERENCES 1. Ahmed, W., et al. (2025). Machine learning-based academic performance prediction. Scientific Reports, 15(1), 12353. 2. Issah, I. (2023). A systematic review of the literature on machine learning in student performance prediction. Education and Information Technologies, 28(4), 1234–1256. 3. Ahmed, E., et al. (2024). Student performance prediction using machine learning: A comparative study. Computers in Education, 78, 123–135. 4. Wang, J., & Yu, Y. (2025). Machine learning approach to student performance prediction of online learning. PLOS ONE, 20(1), e0299018. 5. Villegas, W. (2025). Machine learning models for academic performance prediction: A scalable and interpretable approach. Frontiers in Education, 10, 1632315. 6. Al-Tameemi, G., et al. (2024). A hybrid machine learning approach for predicting student academic performance. Procedia Computer Science, 185, 144–151. 7. Tarik, A. (2021). Artificial intelligence and machine learning to predict student performance: A case study in Morocco. Journal of Educational Technology & Society, 24(4), 112– 124. 8. Albreiki, B. (2021). A systematic literature review of student performance prediction models. Education Sciences, 11(9), 552. 9. Luo, Z., et al. (2024). A method for prediction and analysis of student performance incorporating multidimensional spatiotemporal features. Mathematics, 12(22), 3597. 10. Chen, Y., et al. (2025). Machine learning-driven student performance prediction and tiered instruction integration. arXiv preprint arXiv:2502.03143.