Ho Chi Minh City Open University Journal of Science: Engineering and Technology

Dynamic matrix factorization-based collaborative filtering in movie recommendation services

2024-03-22T10:36:06+07:00

Movies are a primary source of entertainment, but finding specific content can be challenging given the exponentially increasing number of movies produced each year. Recommendation systems are extremely useful for solving this problem. While various approaches exist, Collaborative Filtering (CF) is the most straightforward. CF leverages user input and historical preferences to determine user similarity and suggest movies. Matrix Factorization (MF) is one of the most popular Collaborative Filtering (CF) techniques. It maps users and items into a joint latent space, using a vector of latent features to represent each user or item. However, traditional MF techniques are static, while user cognition and product variety are constantly evolving. As a result, traditional MF approaches struggle to accommodate the dynamic nature of user-item interactions. To address this challenge, we propose a Dynamic Matrix Factorization CF model for movie recommendation systems (DMF-CF) that considers the dynamic changes in user interactions. To validate our approach, we conducted evaluations using the standard MovieLens dataset and compared it to state-of-the-art models. Our preliminary findings highlight the substantial benefits of DMF-CF, which outperforms recent models on the MovieLens-100K and MovieLens-1M datasets in terms of Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) metrics.

A sum rate maximization problem in uplink MIMO with RSMA systems

2024-03-22T10:36:07+07:00

This study explores the problem of maximizing the sum rate in uplink multi-user Multiple-Input Multiple-Output (MIMO) using Rate-Splitting Multiple Access (RSMA) systems. The investigation revolves around the scenario where the Users (UEs) are single-antenna nodes transmitting data to a multi-antenna Base Station (BS) through the RSMA technique. The optimization process encompasses determining parameters such as UEs’ transmit powers, decoding order, and detection vector at the BS. An approach based on Deep Reinforcement Learning (DRL) is introduced to address this challenge. This DRL framework involves an action-refined stage and applies a Deep Deterministic Policy Gradient (DDPG)-based strategy. Simulation outcomes effectively demonstrate the convergence of the proposed DRL framework, where it converges after approximately 1,800 episodes. Also, the results prove the superior performance of the proposed method when compared to established benchmark strategies, where it is up to 45% and 86% higher than the local search and random schemes, respectively.

Integrating BIM and computer vision for preventing Hazards at construction sites

2024-03-22T10:36:07+07:00

Construction safety monitoring is vital in enhancing site safety, such as tracking entering hazardous areas and the correlation between workers and other hazard entities. Therein, computer vision-based image/video processing, one of the emerging technologies, has been actively used to automatically identify and recognize unsafe conditions. However, the construction site has various potential hazard situations during the project. Due to the site’s complexity, many visual devices simultaneously participate in monitoring. It challenges developing and operating corresponding detection algorithms at specific workplaces and times. Besides, safety information detected by computer vision must be organized before being delivered to stakeholders. Hence, this study proposes an approach for construction safety monitoring using vision intelligence technology and BIM-cloud, called BMT. The BMT comprises two modules: (1) the virtual model based on the 4D BIM-cloud model, which provides the spatial-temporal information to decide computer vision algorithm adoptions; (2) the construction physical model built the vision intelligence technologies, which is supported by (1) and deliver safety status and update into the BIM-cloud model to visualize and deliver the risk level to related employees. The efficiency of the BMT approach is validated by testing with the preliminary implementation of a prototype.

Detecting spelling errors in Vietnamese administrative document using large language models

2024-03-22T10:36:08+07:00

In the context of the emergence of more and more administrative documents, the need to ensure accuracy and improve the quality of these documents becomes increasingly important. This research focuses on applying advanced language models to detect spelling errors in administrative documents. Specifically, in this study, a new method using a language model based on the Transformers architecture is proposed to automatically detect and correct common spelling errors in administrative documents. This method combines the model’s ability to understand context and grammar to identify words or phrases that are likely to be misspelled. The proposed method is tested on a dataset containing real administrative documents, and the experimental results show that the proposed model is capable of detecting spelling errors with significant performance, helping to improve accuracy. and improve the quality of administrative documents. This research not only contributes to improving the quality of administrative documents but also opens up new research directions in applying language models to issues related to natural language processing in the field of administration.

Hybrid knowledge-infused collaborative filtering for enhanced movie clustering and recommendation

2024-03-22T10:36:09+07:00

This article proposes an enhanced knowledge-based collaborative filtering model for movie recommendation services to address the limitations of collaborative filtering in capturing the diverse preferences and specific characteristics of movies. The proposed model integrates external knowledge sources, such as movie plots and reviews, to enrich the recommendation process. By leveraging this additional information, the model can better understand movies’ unique features and attributes, improving recommendation accuracy and relevance. The knowledge-based features are extracted and incorporated into the collaborative filtering framework, enhancing the model’s ability to match user preferences with movie characteristics. Experiments are conducted using the MovieLens dataset to evaluate the proposed model. The MAE and RMSE metrics are employed to assess the quality of recommendations. Comparative analyses are conducted against various baseline approaches, including popularity-based, CF-based, content-based, and hybrid recommendation models. The experimental results demonstrate the effectiveness of the proposed knowledge-based collaborative filtering model. The proposed model consistently outperforms the baselines, providing more accurate and personalized recommendations.

A study on constructing an efficient examination scheduling system

2024-03-22T10:36:10+07:00

The objective of this study is to investigate the final exam scheduling process at the Ho Chi Minh City Open University and develop an automated exam scheduling application. Our primary objectives are to prevent students from having conflicting exam schedules and to ensure that no student has to take more than two exams on the same day. This research focuses on applying graph coloring algorithms to the problem of automatic exam scheduling. Our research findings indicate that the graph coloring algorithm is highly effective for automated exam scheduling. This study has the potential to expand and support the development of an automatic exam scheduling and management system, in line with our overall goals. We conduct the experiments on the practical data at HCMCOU and obtain promising results.

Performance comparison ensemble classifier’s performance in answering frequently asked questions about psychology

2024-03-22T10:36:10+07:00

In today’s era of digital healthcare transformation, there is a growing demand for swift responses to mental health queries. To meet this need, we introduce an AI-driven chatbot system designed to automatically address frequently asked questions in psychology. Leveraging a range of classifiers including Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Naïve Bayes, our system extracts insights from expert data sources and employs natural language processing techniques like LDA Topic Modeling and Cosine similarity to generate contextually relevant responses. Through rigorous experimentation, we find that SVM surpasses Naïve Bayes and KNN in accuracy, precision, recall, and F1-score, making it our top choice for constructing the final response system. This research underscores the effectiveness of ensemble classifiers, particularly SVM, in providing accurate and valuable information to enhance mental health support in response to common psychological inquiries.