Vietnam Journal of Computer Science

Healthcare Recommendation Systems (HRSs) primarily aim to offer advice, recommendations, or suggestions related to human healthcare. Similar to other information systems, datasets affect HRSs’ efficiency. The larger datasets can make the information more diverse and complete. Therefore, the recommendation systems can be more accurate and deliver better performance. In addition, several recent studies have revealed that to enhance the accuracy of systems, there should be a switch from a model-centered method to data-centricity. Therefore, several datasets have been introduced over the past few years to assess the effectiveness of HRSs. However, building new datasets and selecting suitable datasets to evaluate HRSs is an issue. Another problem is that there are presently no articles surveying the datasets that have been published to offer the requirements that a dataset uses to judge the HRSs needs to have. These are the reasons that motivated us to conduct this research. To address these problems, we systematically focus on a comprehensive survey of the datasets used to evaluate HRSs published in recent years. Some of the primary contributions of this study are as follows: (i) Surveying 34 datasets commonly used in methods for building and developing HRSs. (ii) Extracting 10 common characteristics and comparing datasets with those 10 common characteristics. (iii) Providing the 13 requirements regarding a dataset used in HRSs (iv) Comparing surveyed datasets and state-of-the-art HRSs based on 13 constructed requirements (v) Discussing the five main challenges and set out some directions for building new datasets in the future. The results of this research analysis are valuable to many new researchers in choosing or constructing suitable datasets to evaluate how they build their HRSs for the present and future.

Prostate cancer is increasingly common among men. However, the process of diagnosing malignant disease is relatively complicated and time-consuming. Identifying benign or malignant tumors early can assist medical professionals in choosing appropriate treatment methods. Consequently, we introduce a soft-voting ensemble model comprising several single machine learning models such as Logistic Regression, Random Forest, XGBoost, LGBM, and Support Vector Machine for the classification task with the prostate cancer dataset. The dataset was divided into two parts for training and testing with a ratio of 67:33. The confusion matrix was used to evaluate the performance of both the individual and ensemble models. Experimental results show that ensemble models achieve performance ranging from 87.88% to 96.97%, which is 3% to 9% better than individual models, surpassing recent research. Integrating the strengths of individual models helps minimize errors, resulting in optimal classification with high accuracy and overall performance in the field of machine learning.

The security of digital communication and information systems is mostly dependent on block ciphers. ARX-based ciphers are widely used due to their simplicity and efficiency. This paper provides an exhaustive cryptanalysis of a subset of ARX-based block ciphers, with particular emphasis on SIMON, SPECK, and IDEA. These ciphers need to be exposed for their weaknesses in algebraic attack resistance and cryptographic properties such as key sensitivity. In addition, we assess the resource utilization and speed of these ciphers, both of which are critical for practical implementation. SMT (Satisfiability Modulo Theories) framework is utilized to tackle constraint fulfillment problems based on first-order logic. The following cryptographic steps use SMT solvers: differential cryptanalysis, collision attack, pre-image attack, modular root-finding, and cryptographic primitive verification. We show that SMT solvers can solve block cipher cryptanalysis constraints. In a cryptanalytic attack, we convert block cipher boolean equations to Z3py. The proposed cryptanalysis method evaluates ARX cipher performance. This method recovers the partial secret key using plaintext and ciphertext pairs, partial key bits, and a predetermined number of rounds. To determine whether SIMON, SPECK, or IDEA are appropriate for distinct security requirements, we conducted a comparative analysis of the results and presented them in tabulated form. This research builds a better understanding of ARX-based block ciphers and allows us to develop more robust and efficient cryptographic algorithms to protect sensitive data in modern communication systems.

Key information extraction and recognition from rich text images are crucial for various applications. There are two main tasks involved in this process: Line Item Recognition (LIR) and Key Information Localization and Extraction (KILE). LIR aims at identifying and interpreting data line items in a document. The essential information in each line item is then classified or extracted, a task known as KILE. A widely used approach for this problem is sequence based, which relies on the generalization of a language model and requires a significant amount of training time. We present an effective and reliable solution to the problem by using RoBERTa, a transformer model trained on a large corpus, along with the LION optimizer to improve the training process. A comprehensive evaluation was conducted on two different benchmarks, emphasizing two different languages, English and Vietnamese. Experimental results on DocILE indicate that the proposed framework significantly improves the KILE task with a 7.24% increase in accuracy compared to the baseline and also enhances the correct recognition rate at the LIR stage. On MCOCR, the method achieved a Character Error Rate (CER) of 28.6%, which is competitive with the state-of-the-art on this dataset.

Due to the diversity of image sources, content-based multi-source image fusion and retrieval have shown promising capabilities in computer vision tasks, and especially when applied in Computer-Aided Diagnosis (CAD) to automate and improve the accuracy of medical image analysis. The combination of computer vision and CAD systems has the potential to revolutionize healthcare by augmenting the expertise of clinicians, improving overall diagnostic accuracy and helping experts in the clinical decision-making process by classifying and retrieving similar annotated clinical images to a given query. In the context of multi-view mammography interpretation, the concept of multi-view feature fusion has recently been studied to improve retrieval performance while effectively guaranteeing the complementarity of both MLO and CC views. However, conventional multi-view feature fusion makes descriptors long and lacks to take into consideration the relationship between descriptors. To deal with this issue, we propose two hierarchical multi-view feature fusion methods, for multi-view mammogram retrieval, based on the Canonical Correlation Analysis (CCA), which is the most commonly used multivariate parametric test. In fact, we have adapted CCA to determine the relationship between two descriptors by processing latent correlation factors. Moreover, after extracting descriptors for each view, a comparative study of texture and shape fusion descriptors is proposed in order to identify the more discriminative features for multi-view mammogram retrieval. Then, a query-dependent distance metric preserving both visual resemblance and semantic similarity is carried out to dynamically determine the more appropriate distance measure for each query image. Extensive experiments on the challenging Curated Breast Imaging Subset of Digital Database for Screening Mammography (CBIS-DDSM) have demonstrated the effectiveness of the proposed hierarchical multi-view feature fusion for mammogram retrieval, which outperforms the performance achieved either by conventional fused information or by single view information. To improve the transparency of our paper, the source code of the proposed method and the related dataset (including readme files) are publicly accessible through the following GitHub link: https://github.com/ABDERRAHIMMAR/Multi-View-Feature-Fusion-for-Mammogram-Retrieval. This open-access resource empowers researchers and practitioners to delve deeper into our methodology, fostering collaboration and advancements in the field of computer-aided diagnosis and medical image analysis.

As the complexity of software systems is ever increasing, so is the need for practical tools for formal verification. Among these are automatic theorem provers, capable of solving various reasoning problems automatically, and proof assistants, capable of deriving more complex results when guided by a mathematician/programmer. In this paper we consider using the latter to build the former. In the proof assistant Isabelle/HOL we combine functional programming and logical program verification to build a theorem prover for propositional logic. We also consider how such a prover can be used to solve a reasoning task without much mental labor. The development is extended with a formalized proof system for writing machine-checked sequent calculus proofs. We consider how this can be used to teach computer science students about logic, automated reasoning and proof assistants.