Journal Article

A Recommender System for Personalized Reading Recommendations and Literature Discovery utilizing the HGRNN-EOO technique

Zhang, Yongjun

2024

Overview

Recommender systems are used to address information overload, enhance personalization, and improve user experience by providing tailored suggestions based on individual preferences, thereby increasing engagement and facilitating content discovery. This paper proposes a hybrid approach for recommender system in personalized reading recommendation and literature discovery. The proposed hybrid approach is the combined performance of both the Hierarchal Gated Recurrent Neural Network (HGRNN) and Eurasian Oystercatcher Optimizer (EOO). Commonly it is named as HGRNN-EOO technique. The major objective of the proposed approach is to provide a recommender system for personalized reading recommendation and literature discovery. HGRNN is designed to provide personalized recommendations based on their preferences, behaviour, and interactions to enhance user experience and engagement. The personalized recommendations from the HGRNN are optimized by using the EOO. By then, the MATLAB working platform has been proposed and implemented, and the present processes are used to calculate the execution. Using performance metrics like accuracy, error rate, F-score, precision, recall, computation time, ROC, sensitivity, and specificity, the proposed method's effectiveness is evaluated. From the result, the proposed approach based error is less compared to existing techniques. The result shows that the accuracy level of proposed Recommender System in Personalized Reading Recommendation using Hierarchal Gated Recurrent Neural Network and Eurasian Oystercatcher Optimizer (RSPRR-HGRNN-EOO) approach is 98% that is higher than the other existing methods. The specificity and the F-score of the proposed RSPRR-HGRNN-EOO approach is 99% and 97%. The error rate of the proposed RSPRR-HGRNN-EOO approach is 2.5%, which is very less compared to other existing techniques. The proposed method shows better results in all existing methods like Recommender System in Personalized Reading Recommendation Convolutional Neural Network (RSPRR-CNN), Recommender System in Personalized Reading Recommendation Deep Neural Network (RSPRR-DNN) and Recommender System in Personalized Reading Recommendation Feed-Forward Neural Network (RSPRR-FNN). Based on the outcome, it can be concluded that the proposed strategy has a lower error rate than existing methods.

• Publisher: Engineering and Scientific Research Groups
• Subject: Artificial neural networks / Customization / Hybrid systems / Neural networks / Performance measurement / Reading / Recommender systems / Recurrent neural networks / User experience