Article
Computer Science, Theory & Methods
Zhenyan Ji, Mengdan Wu, Hong Yang, Jose Enrique Armendariz Inigo
Summary: This paper proposes a time-sensitive heterogeneous graph neural network for news recommendation, where one subnet learns the temporal characteristics of user behavior and the other subnet builds an attention-based graph to model user-news-topic associations. Experiments show that the proposed model outperforms state-of-the-art models in accuracy and interpretability.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2021)
Article
Computer Science, Theory & Methods
Imran Ahmed, Misbah Ahmad, Abdellah Chehri, Gwanggil Jeon
Summary: In the healthcare sector, patient data is crucial for medical diagnoses and treatment plans. Existing techniques for finding similar patients based on Electronic Health Record (EHR) data face challenges due to high-dimensional and sparse vectors. To overcome this, the paper proposes a novel heterogeneous network-embedded drug recommendation system. The system focuses on classifying the sentiment of drug users based on their reviews and relevant features, achieving a classification accuracy of 92%.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2023)
Article
Computer Science, Software Engineering
Xin Zhang, Yan Yang
Summary: This research proposes a heterogeneous graph network that combines the idea of a bilayer-attention network with the self-attention mechanism for recommendation algorithms. The proposed algorithm, HARec, improves recommendation accuracy and captures more abundant information, showing good interpretability.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2022)
Article
Computer Science, Information Systems
Qing Meng, Hui Yan, Bo Liu, Xiangguo Sun, Mingrui Hu, Jiuxin Cao
Summary: In news recommendation, the sparsity of user behavior data due to the overwhelming amount of news daily poses a challenge for providing personalized recommendations. This study addresses this issue by integrating user-news relationships and overall user historical clicked news sequences to construct a global heterogeneous transition graph. A refinement approach is used to identify news transition patterns in the graph. Additionally, a heterogeneous transition graph attention network is proposed to capture common behavior patterns among users and enhance user interest representation. The proposed GAINREC model effectively combines personalized and common user interests for news recommendation, outperforming existing models according to experimental results.
ACM TRANSACTIONS ON INFORMATION SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Danyang Wang, Xi Xiong, Yuanyuan Li, Jianghe Wang, Qiurong Tan
Summary: Matching candidate news with user interests is critical for news recommendation. To confront the challenge of multiple user interests, a hierarchical candidate-aware user modeling framework is proposed to accurately match users' multi-field and multi-grained interests with candidate news. Experimental outcomes on large-scale datasets demonstrate the effectiveness and superiority of the proposed method over existing state-of-the-art methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Wenting Zhang
Summary: This article develops a news recommendation model based on a sub-attention news encoder to extract finer-grained segment features from news and represent users accurately and exhaustively. The model utilizes CNN and sub-attention mechanism to extract a rich feature matrix from the news text. It also incorporates a multi-head self-attention mechanism and time series prediction for the user's interests. Experimental results demonstrate that the proposed model performs well on various indicators and outperforms other models in terms of convergence speed, providing guidance for future news recommendation system designs.
PEERJ COMPUTER SCIENCE
(2023)
Article
Mathematical & Computational Biology
Suqi Zhang, Xinxin Wang, Wenfeng Wang, Ningjing Zhang, Yunhao Fang, Jianxin Li
Summary: A recommendation model based on intention decomposition and heterogeneous information fusion (IDHIF) is proposed in this study to enhance the accuracy and effectiveness of recommendations. The model decomposes user and item interaction intentions, mines short feature representations, and utilizes heterogeneous information fusion techniques to mine interactive, social, and content features. Experimental results show that the proposed model outperforms the baseline model in terms of AUC, F1, and Recall@20, demonstrating its ability to better model user and item features and improve recommendation efficiency.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Heng-Shiou Sheu, Zhixuan Chu, Daiqing Qi, Sheng Li
Summary: Personalized news recommendation aims to recommend news articles based on user preferences and short-term reading interests, and session-based news recommendation has recently attracted attention for recommending news articles within an active session. The CAGE approach improves semantic representations of news articles using external knowledge graphs, graph neural networks, and attention neural networks to model user preferences, outperforming competitive baselines in multiple news recommendation benchmark datasets.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Green & Sustainable Science & Technology
Surya Michrandi Nasution, Emir Husni, Kuspriyanto Kuspriyanto, Rahadian Yusuf, Bernardo Nugroho Yahya
Summary: The paper proposes a framework for a contextual route recommendation system that is compatible with traffic conditions and vehicle type, along with other relevant attributes. The framework consists of two phases: traffic prediction using Knowledge-Growing Bayes Classifier and route recommendation using a new measure called road capacity value along with the Dijkstra algorithm. The performance of traffic prediction is around 60.78-73.69% for accuracy, 63.64-77.39% for precision, and 60.78-73.69% for recall.
Article
Political Science
Philipp M. Lutscher
Summary: Most authoritarian countries censor the press, leading many opposition and independent news outlets to seek refuge on the internet. This study examines Denial-of-Service attacks on news websites in Venezuela and finds a correlation between news content and these attacks.
POLITICAL SCIENCE RESEARCH AND METHODS
(2021)
Article
Computer Science, Information Systems
Md Shafiul Alam Forhad, Mohammad Shamsul Arefin, A. S. M. Kayes, Khandakar Ahmed, Mohammad Jabed Morshed Chowdhury, Indika Kumara
Summary: This paper introduces a framework that ranks hotels by analyzing customer reviews and nearby amenities, as well as a framework that combines user reviews and surrounding facility scores. Experimental results confirm the effectiveness of the proposed recommendation framework.
Article
Computer Science, Artificial Intelligence
Sourya Dipta Das, Ayan Basak, Saikat Dutta
Summary: The significance of social media has greatly increased as it allows people from all over the world to stay connected, but this also leads to the circulation of fake news and tweets. In this paper, a novel Fake News Detection system using pre-trained models and statistical feature fusion network is proposed to automatically identify real and fake news. The system shows effectiveness in detecting fake news in short news content as well as in news articles on different datasets.
Article
Computer Science, Artificial Intelligence
Ala Mughaid, Shadi Al-Zu'bi, Ahmed Al Arjan, Rula AL-Amrat, Rathaa Alajmi, Raed Abu Zitar, Laith Abualigah
Summary: This paper explores the technique of automatically detecting fake news and proposes a method that uses the world rank of news websites as the main factor of news accuracy and compares current news with fake news to determine their accuracy. Experimental results show that the proposed method performs well in defining news accuracy.
Article
Computer Science, Information Systems
Safia Kanwal, Muhammad Kamran Malik, Zubair Nawaz, Khawar Mehmood
Summary: The study proposes an Urdu entity linking pipeline and prepares a relevant dataset. The pipeline and dataset are used to perform entity linking and annotation on Urdu news articles. A subknowledge graph is extracted and used to build an Urdu news recommendation system, achieving an accuracy of 60.8%.
Article
Physics, Multidisciplinary
R. Latha
Summary: Collaborative Filtering approaches are essential tools for recommending products based on historical knowledge. This study proposes User Trait Model and Bayesian Global Agreement model to improve recommendation quality in low correlated data. The proposed approaches outperform other user based CF approaches in terms of accuracy and diversity of recommendations.
PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS
(2022)