Stack Overflow is a popular Community-based Question Answer (CQA) website focused on software programming and has attracted more and more users in recent years. However, duplicate questions frequently appear in Stack Overflow and they are manually marked by the users with high reputation. Automatic duplicate question detection alleviates labor and effort for users with high reputation. Although existing approaches extract textual features to automatically detect duplicate questions, these approaches are limited since semantic information could be lost. To tackle this problem, we explore the use of powerful deep learning techniques, including Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM), to detect duplicate questions in Stack Overflow. In addition, we use Word2Vec to obtain the vector representations of words. They can fully capture semantic information at document-level and word-level respectively. Therefore, we construct three deep learning approaches WV-CNN, WV-RNN and WV-LSTM, which are based on Word2Vec, CNN, RNN and LSTM, to detect duplicate questions in Stack Overflow. Evaluation results show that WV-CNN and WV-LSTM have made significant improvements over four baseline approaches (i.e., DupPredictor, Dupe, DupPredictorRep-T, and DupeRep) and three deep learning approaches (i.e., DQ-CNN, DQ-RNN, and DQ-LSTM) in terms of recall-rate@5, recall-rate@10 and recall-rate@20. Furthermore, the experimental results indicate that our approaches WV-CNN, WV-RNN, and WV-LSTM outperform four machine learning approaches based on Support Vector Machine, Logic Regression, Random Forest and eXtreme Gradient Boosting in terms of recall-rate@5, recall-rate@10 and recall-rate@20.

Detecting and refactoring code smells is challenging, laborious, and sustaining. Although large language models have demonstrated potential in identifying various types of code smells, they also have limitations such as input-output token restrictions, difficulty in accessing repository-level knowledge, and performing dynamic source code analysis. Existing learning-based methods or commercial expert toolsets have advantages in handling complex smells. They can analyze project structures and contextual information in-depth, access global code repositories, and utilize advanced code analysis techniques. However, these toolsets are often designed for specific types and patterns of code smells and can only address fixed smells, lacking flexibility and scalability. To resolve that problem, we propose iSMELL, an ensemble approach that employs various code smell detection toolsets via Mixture of Experts (MoE) architecture for comprehensive code smell detection, and enhances the LLMs with the detection results from expert toolsets for refactoring those identified code smells. First, we train a MoE model that, based on input code vectors, outputs the most suitable expert tool for identifying each type of smell. Then, we select the recommended toolsets for code smell detection and obtain their results. Finally, we equip the prompts with the detection results from the expert toolsets, thereby enhancing the refactoring capability of LLMs for code with existing smells, enabling them to provide different solutions based on the type of smell. We evaluate our approach on detecting and refactoring three classical and complex code smells, i.e., Refused Bequest, God Class, and Feature Envy. The results show that, by adopting seven expert code smell toolsets, iSMELL achieved an average F1 score of 75.17% on code smell detection, outperforming LLMs baselines by an increase of 35.05% in F1 score. We further evaluate the code refactored by the enhanced LLM. The quantitative and human evaluation results show that iSMELL could improve code quality metrics and conduct satisfactory refactoring toward the identified code smells. We believe that our proposed solution could provide new insights into better leveraging LLMs and existing approaches to resolving complex software tasks.