Machine Learning Based Methods for Arabic Duplicate Question Detection

In this work, a best answer recommendation model is proposed for a Question Answering (QA) system. A Community Question Answering System was subsequently developed based on the model. The system applies Brouwer Fixed Point Theorem to prove the existence of the desired voter scoring function and Normalized Google Distance (NGD) to show closeness between words before an answer is suggested to users. Answers are ranked according to their Fixed-Point Score (FPS) for each question. Thereafter, the highest scored answer is chosen as the FPS Best Answer (BA). For each question asked by user, the system applies NGD to check if similar or related questions with the best answer had been asked and stored in the database. When similar or related questions with the best answer are not found in the database, Brouwer Fixed point is used to calculate the best answer from the pool of answers on a question then the best answer is stored in the NGD data-table for recommendation purpose. The system was implemented using PHP scripting language, MySQL for database management, JQuery, and Apache. The system was evaluated using standard metrics: Reciprocal Rank, Mean Reciprocal Rank (MRR) and Discounted Cumulative Gain (DCG). The system eliminated longer waiting time faced by askers in a community question answering system. The developed system can be used for research and learning purposes.

User votes are important signals in community question-answering (CQA) systems. Many features of typical CQA systems, e.g. the best answer to a question, status of a user, are dependent on ratings or votes cast by the community. In a popular CQA site, Yahoo! Answers, users vote for the best answers to their questions and can also thumb up or down each individual answer. Prior work has shown that these votes provide useful predictors for content quality and user expertise, where each vote is usually assumed to carry the same weight as others. In this paper, we analyze a set of possible factors that indicate bias in user voting behavior -- these factors encompass different gaming behavior, as well as other eccentricities, e.g., votes to show appreciation of answerers. These observations suggest that votes need to be calibrated before being used to identify good answers or experts. To address this problem, we propose a general machine learning framework to calibrate such votes. Through extensive experiments based on an editorially judged CQA dataset, we show that our supervised learning method of content-agnostic vote calibration can significantly improve the performance of answer ranking and expert ranking.

The task of finding- and ranking-related questions plays the most important role for any real-world Community Question Answering (cQA) systems. This paper proposes a new method to solve this problem by considering multi-views for measuring the similarities between the input questions and the question-answering pairs in the database. Our model will investigate various aspects for understanding questions. Beside the traditional features such as bag of n-grams, we will use more efficient aspects that include word embeddings and question categories. We will use a word representation model for generating word embeddings, a question classification module for determining the category for an input question. Then all these obtained features are combined into a machine learning-based framework for getting similarity existing question-answering pairs as well as for ranking these pairs. We tested our proposed approach on the dataset SemEval 2016 and the experiment shows obtained results with the Accuracy and MAP of 80.43% and 77.43%, respectively, which are the highest accuracies in comparison with previous studies.

This paper presents an extension of the community question-answering (CQA) system we developed previously. This graph-based system method that builds ranked answers for related questions using nKullback–Leibler (KL) divergence. The process of extracting answers to questions in this work involves; question core, building question query, query-based answer extraction (QBAE), pattern-based answer extraction (PBAE), and combined answer extraction. The source data for this work were existing data from ResearchGate, a socio-academic networking website that provides researchers the platform to collaborate, ask questions, and offer answers to questions. The performance for answer extraction for 2786 questions shows that when 80% of patterns and keywords were considered, QBAE and PBAE extracted 2765 and 2766 correct answers, respectively, while the QBAE + PBAE method extracted 2782 correct answers. Also, when 90% of patterns and keywords were utilized, QBAE and PBAE extracted 2782 and 2784 correct answers, whereas the QBAE + PBAE method extracted 2786 correct answers. Our method was able to identify 229 questions without answers. Finally, the evaluation of our model reveals high-performance accuracy and precision.

Community question answering (CQA) systems aim to provide users with high-quality answers. Nevertheless, unreliable answers are often returned to users in CQA systems, and the phenomenon causes that users have to browse multiple answers to find the best one. To improve such problem, we design a novel scheme, named PW-CrossCMN. The scheme ranks the candidate answers by pair-wise approach based on numerous historical documents. In the scheme, we apply the preference relationship into deep learning framework. Specifically, the scheme consists of two phases. In phase 1, the scheme extracts the features via automated feature engineering to construct the preference vectors and then divides the vectors into balanced positive and negative training samples based on the preference relationship. In phase 2, we build the CrossCMN model, which implements the multi-network parallel convolution and the cross forward propagation of full-connected layers, to achieve training and prediction tasks. Moreover, the multi-layer perception (MLP) is introduced to extract combination features in the prediction module. We perform extensive experiments on two typical datasets, and the results show that our scheme has more excellent performance in answer ranking task compared with several state-of-the-art baselines. In addition, we have released the relevant codes.

Intelligent Community Question Answering (CQA) system is a popular research topic in Natural Language Processing (NLP) area. Recently, the development of distributed data mining has significantly improved the efficiency of the searching and sorting operations in CQA database. However, problems are left unanswered such as how to choose the best answer among multiple candidates for a single question. To solve this issue, we design a CQA system called UB-CQA. First, in database refinement part, we construct an optimized and structured database by the best answer choosing method leveraging user attribute provided by the response provider. Second, in human-computer part, the UB-CQA is able to search and provide a more satisfying answer to users by similarity calculation and re-ranking method leveraging text categorization information. Empirical evaluations show that the best answer choosing and the candidate question re-ranking methods bring great improvements in accuracy and reliability.

Community question answering (cQA) systems have enjoyed the benefits of advances in neural information retrieval, some models of which need annotated documents as supervised data. However, in contrast with the amount of supervised data for cQA systems, user-generated data in cQA sites have been increasing greatly with time. Thus, focusing on unsupervised models, we tackle a task of retrieving relevant answers for new questions from existing cQA data and propose two frameworks to exploit a Question Retrieval (QR) model for Answer Retrieval (AR). The first framework ranks answers according to the combined scores of QR and AR models and the second framework ranks answers using the scores of a QR model and best answer flags. In our experiments, we applied the combination of our proposed frameworks and a classical fusion technique to AR models with a Japanese cQA data set containing approximately 9.4M question-answer pairs. When best answer flags in the cQA data cannot be utilized, our combination of AR and QR scores with data fusion outperforms a base AR model on average. When best answer flags can be utilized, the retrieval performance can be improved further. While our results lack statistical significance, we discuss effect sizes as well as future sample sizes to attain sufficient statistical power.

This paper proposes an automated question answering system based on improved Term Frequency- Inverse Document Frequency (TF-IDF) and cosine similarity. The main purpose of this research study is to provide an effective question answering system that retrieves precise and relevant answers to the users' queries with high confidence. The existing studies in the relevant literature show that several techniques have been proposed for automatic question answering systems. The rule-based techniques depend on inference rules and take less time to respond to the user query. However, generations of pattern and inference rules are difficult as natural languages lack to follow a fixed pattern. The content-based similarity method pre-computes similarity with all the repository questions for a given query. In this research study, firstly all repository questions are pre-processed and a matrix using the improved TF -IDF model is generated. Then, we find the similarity of each user query with the matrix after query pre-processing. For the proposed approach, we remove stop-words and apply lemmatization and POS tagging techniques for pre-processing. The proposed framework is implemented using the standard datasets used in the existing studies. The empirical analysis-based results show that the systems adopting the proposed technique takes less than five seconds to respond to user queries with maximum similarity. The proposed framework attains up to 84% accuracy.

There have been increasing interests in Community Question Answering (CQA) recently. CQA websites such as Yahoo! Answers and Baidu Knows are increasingly popular, attracting tens of thousands of users to submit questions and answers every day. However, we find that there is a gap in the study of what kinds of questions are more likely to attract answers, which makes sense for users in practice when asking questions in the CQA systems. In this paper, we investigate the factors that may affect users’ questions to attract answers. We also introduce a general framework to predict how many replies a question is expected to receive. Evaluation results of the framework on real data prove its effectiveness.

Text classifications have been popular in recent years. To classify the text, the first step that needs to be done is to convert the text into some value. Some values that can be used, such as Term Frequencies, Inverse Document Frequencies, Term Frequencies – Inverse Document Frequencies, and Frequency of the word itself. This study aims to get which metric value is best in text classification. The method used is Naïve Bayes, Logistic Regression, and Random Forest. The evaluation score that is used is accuracy and Area Under Curve value. It comes out that some metric values produce similar evaluation scores. Another finding is that Random Forest is the best method among others, also the best metric for text classification is Term Frequencies – Inverse Document Frequencies.

Automatic assessment of concept map quality is an important challenge in the field of education, particularly in evaluating students' conceptual understanding objectively and efficiently. This study aims to compare the performance of two machine learning algorithms, namely Random Forest and Ordinal Regression, in classifying the quality of concept maps. The evaluation was conducted on three approaches to text feature representation: Term Frequency-Inverse Document Frequency (TF-IDF), Bidirectional Encoder Representations from Transformers (BERT), and a combination of both (TF-IDF + BERT). Additionally, this study compares the performance of the models under two dataset conditions: original data and data balanced using the Synthetic Minority Over-sampling Technique (SMOTE), to address the class imbalance that often occurs in educational data. The data used consists of a collection of propositions from students' concept maps that have been labeled with ordinal scores based on quality. Text representation is extracted using the TF-IDF and BERT approaches, and then used as input to build the classification model. Performance evaluation was conducted using the metrics of Accuracy, Precision, Recall, F1-score, Cohen’s Kappa, and MAE. The results show that the Ordinal Regression model with TF-IDF representation combined with SMOTE achieved the best performance, with an accuracy of 0.8777, an F1-score of 0.8773, and a Cohen’s Kappa of 0.7701. These results indicate that classical feature representations like TF-IDF remain effective in limited data scenarios, and that the SMOTE technique successfully improved the model's performance by reducing bias towards the majority class. This research contributes to the development of an automatic concept map assessment system and suggests optimal classification strategies for educational datasets with ordinal and imbalanced characteristics

The rapid growth of financial technology (fintech) services in Indonesia has significantly transformed digital transaction behavior, with digital wallets such as DANA becoming widely adopted. Despite high usage rates, public discourse on social media frequently highlights both positive experiences and recurring concerns related to system reliability and data security. Twitter, as a real-time public communication platform, provides a rich source of user-generated content that reflects public perception toward fintech services. This study investigates public sentiment toward DANA fintech services in Indonesia through sentiment analysis of Twitter data using machine learning approaches. Purpose:This study aims to analyze public perception of DANA fintech services based on sentiment expressed on Twitter and to compare the performance of several machine learning classification algorithms in identifying positive, neutral, and negative sentiments in Indonesian-language social media texts. Methods/Study design/approach: The study employs a quantitative descriptive approach using text mining and sentiment analysis techniques. Twitter data were collected via the Twitter API over a specified period and processed through preprocessing stages including text cleaning, tokenization, stopword removal, and stemming. Feature extraction was performed using the Term Frequency–Inverse Document Frequency (TF-IDF) method. Sentiment classification was conducted using Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Random Forest algorithms. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics. Result/Findings:The results indicate that public sentiment toward DANA is predominantly positive, driven by perceptions of transaction convenience and speed, followed by neutral sentiment, while negative sentiment is mainly associated with system disruptions and data security issues. Among the evaluated algorithms, SVM achieved the highest classification accuracy (91.4%), outperforming Random Forest and KNN. These findings demonstrate that SVM is more effective for sentiment classification of short Indonesian-language texts on social media platforms. Novelty/Originality/Value:This study contributes to the growing body of sentiment analysis research in Indonesia by providing empirical evidence on public perception of fintech services using machine learning techniques. By systematically comparing multiple classification algorithms in the context of Indonesian Twitter data, the study offers methodological insights and practical implications for fintech service providers in improving service quality, communication strategies, and user trust.

Android is the most widely used mobile operating system (OS). A large number of third-party Android application (app) markets have emerged. The absence of third-party market regulation has prompted research institutions to propose different malware detection techniques. However, due to improvements of malware itself and Android system, it is difficult to design a detection method that can efficiently and effectively detect malicious apps for a long time. Meanwhile, adopting more features will increase the complexity of the model and the computational cost of the system. Permissions play a vital role in the security of the Android apps. Term Frequency—Inverse Document Frequency (TF-IDF) is used to assess the importance of a word for a file set in a corpus. The static analysis method does not need to run the app. It can efficiently and accurately extract the permissions from an app. Based on this cognition and perspective, in this paper, a new static detection method based on TF-IDF and Machine Learning is proposed. The system permissions are extracted in Android application package’s (Apk’s) manifest file. TF-IDF algorithm is used to calculate the permission value (PV) of each permission and the sensitivity value of apk (SVOA) of each app. The SVOA and the number of the used permissions are learned and tested by machine learning. 6070 benign apps and 9419 malware are used to evaluate the proposed approach. The experiment results show that only use dangerous permissions or the number of used permissions can’t accurately distinguish whether an app is malicious or benign. For malware detection, the proposed approach achieve up to 99.5% accuracy and the learning and training time only needs 0.05s. For malware families detection, the accuracy is 99.6%. The results indicate that the method for unknown/new sample’s detection accuracy is 92.71%. Compared against other state-of-the-art approaches, the proposed approach is more effective by detecting malware and malware families.

IntroductionAccurate classification of brain tumors from MRI scans is a critical task for improving patient outcomes. Machine learning (ML) and deep learning (DL) methods have shown promise in this domain, but their relative performance remains unclear.MethodsThis study evaluates several ML and DL techniques using the BraTS 2024 dataset. The models assessed include traditional algorithms such as Random Forest and advanced deep learning architectures including Simple CNN, VGG16, VGG19, ResNet50, Inception-ResNetV2, and EfficientNet. Preprocessing strategies were applied to optimize model performance.ResultsThe Random Forest classifier achieved the highest accuracy of 87%, outperforming all deep learning models, which achieved accuracy in the range of 47% to 70%. This indicates that traditional ML approaches can sometimes surpass state-of-the-art DL methods in tumor classification tasks.DiscussionThe findings highlight the importance of model selection and parameter tuning in automated brain tumor diagnosis. While deep learning models are generally considered standard for image analysis, Random Forest demonstrated superior performance in this context. This underscores the need for fine-grained consideration of dataset characteristics, computational resources, and diagnostic requirements.ConclusionThe study shows that carefully selected and optimized ML approaches can improve tumor classification and support more accurate and efficient diagnostic systems for brain tumor patients.

Recently, deep learning methods, for example, convolutional neural networks (CNNs), have achieved high performance in hyperspectral image (HSI) classification. The limited training samples of HSI images make it hard to use deep learning methods with many layers and a large number of convolutional kernels as in large scale imagery tasks, and CNN-based methods usually need long training time. In this paper, we present a wide sliding window and subsampling network (WSWS Net) for HSI classification. It is based on layers of transform kernels with sliding windows and subsampling (WSWS). It can be extended in the wide direction to learn both spatial and spectral features more efficiently. The learned features are subsampled to reduce computational loads and to reduce memorization. Thus, layers of WSWS can learn higher level spatial and spectral features efficiently, and the proposed network can be trained easily by only computing linear weights with least squares. The experimental results show that the WSWS Net achieves excellent performance with different hyperspectral remotes sensing datasets compared with other shallow and deep learning methods. The effects of ratio of training samples, the sizes of image patches, and the visualization of features in WSWS layers are presented.