View in EDS

A Hybrid Recommender System for Scientific Articles Using Semantic Analysis.

Saved in:
Bibliographic Details
Title: A Hybrid Recommender System for Scientific Articles Using Semantic Analysis.
Authors: Moukhtar, Basma, Makady, Soha, Ezzat, Cherry Ahmed
Source: International Journal of Intelligent Engineering & Systems; 2026, Vol. 19 Issue 2, p357-383, 27p
Subject Terms: CONTENT analysis, RECOMMENDER systems, PERIODICAL articles, STUDENT interests
Abstract: Researchers spend considerable time filtering through research papers to find relevant resources. This slows down the research process and reduces productivity, especially for junior researchers with less experience. To address this challenge, we propose a Hybrid Recommender System for Scientific Articles (HRSSA) that combines content based, collaborative-based and semantic-based recommendation approaches. By applying topic modeling, we identify topics in our dataset, then each paper in the dataset is labelled with a topic. The recommendation process starts by matching the user’s query to the most relevant topic and all papers associated with that topic are retrieved. Furthermore, topic-based user profiles are created for all papers’ authors, including the user who submitted the query. The identified papers are then ranked based on their topic relevance percentage and similarity to the user profile. Recognizing that research topics are often interrelated, we utilize topic modeling to semantically identify related topics within the same cluster. Additionally, users are grouped based on their research interests to enhance the recommendation process. Our approach not only retrieves related topics but also suggests papers favoured by users with similar interests. This hybrid strategy significantly improves performance metrics on DBLP version 11 dataset, surpassing state-of-the-art methods with a 25.6% increase in Mean Reciprocal Rank (MRR), a 25% increase in Mean Average Precision (MAP), and a 10% increase in Recall. Similarly, for CiteULike dataset, we surpassed the related work by 45.6% in precision, 40% in NDCG and 68% in MRR. Overall, our experimental results demonstrate that the proposed framework effectively enhances the paper recommendation process, helping researchers find resources aligned with their interests. [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Intelligent Engineering & Systems is the property of Intelligent Networks & Systems Society and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Complementary Index
Description
Abstract:Researchers spend considerable time filtering through research papers to find relevant resources. This slows down the research process and reduces productivity, especially for junior researchers with less experience. To address this challenge, we propose a Hybrid Recommender System for Scientific Articles (HRSSA) that combines content based, collaborative-based and semantic-based recommendation approaches. By applying topic modeling, we identify topics in our dataset, then each paper in the dataset is labelled with a topic. The recommendation process starts by matching the user’s query to the most relevant topic and all papers associated with that topic are retrieved. Furthermore, topic-based user profiles are created for all papers’ authors, including the user who submitted the query. The identified papers are then ranked based on their topic relevance percentage and similarity to the user profile. Recognizing that research topics are often interrelated, we utilize topic modeling to semantically identify related topics within the same cluster. Additionally, users are grouped based on their research interests to enhance the recommendation process. Our approach not only retrieves related topics but also suggests papers favoured by users with similar interests. This hybrid strategy significantly improves performance metrics on DBLP version 11 dataset, surpassing state-of-the-art methods with a 25.6% increase in Mean Reciprocal Rank (MRR), a 25% increase in Mean Average Precision (MAP), and a 10% increase in Recall. Similarly, for CiteULike dataset, we surpassed the related work by 45.6% in precision, 40% in NDCG and 68% in MRR. Overall, our experimental results demonstrate that the proposed framework effectively enhances the paper recommendation process, helping researchers find resources aligned with their interests. [ABSTRACT FROM AUTHOR]
ISSN:2185310X
DOI:10.22266/ijies2026.0228.23