Scholarlink

Stack Overflow is a widely used platform for programmers to seek and share knowledge through questions and answers. A programmer asks questions and assigns tags which act as labels for facilitation of easier search and discovery. Based on these questions, the community responds with feasible answers on the website. However, the platform lacks an automated tagging system, relying on manual tagging which can be inconsistent and inefficient sometimes. This paper addresses the limitations of manual tagging by automating the process. The automated tagging framework incorporates advanced machine learning (ML) and deep learning (DL) techniques to enhance the accuracy and efficiency of tag assignments. The data used for this study comes from Hugging Face and Kaggle, which includes a comprehensive collection of Stack Overflow questions and their corresponding tags. We used preprocessing steps such as tokenization, normalization, and padding to prepare the data for training. Traditional ML models like Naive Bayes, Support Vector Machines (SVM), Random Forest, and K-Nearest Neighbors (KNN) were initially explored, yielding an accuracy of around 70%. However, a hybrid model incorporating Gated Recurrent Units (GRU), Convolutional Neural Networks (CNN), and attention mechanisms significantly outperformed these models, providing better accuracy and computational efficiency. The findings of this paper highlight the potential of integrating natural language processing (NLP) techniques to automate the tagging process on platforms like Stack Overflow, improving the organization and searchability of content.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Automating Tag Prediction in Stack Overflow Using Machine Learning and Deep Learning Techniques

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Riya C. Sapkale,
  • Pranav Acharya,
  • Shreya M. Hegde,
  • Shreya Kiran,
  • Ashwini M. Joshi

摘要

Stack Overflow is a widely used platform for programmers to seek and share knowledge through questions and answers. A programmer asks questions and assigns tags which act as labels for facilitation of easier search and discovery. Based on these questions, the community responds with feasible answers on the website. However, the platform lacks an automated tagging system, relying on manual tagging which can be inconsistent and inefficient sometimes. This paper addresses the limitations of manual tagging by automating the process. The automated tagging framework incorporates advanced machine learning (ML) and deep learning (DL) techniques to enhance the accuracy and efficiency of tag assignments. The data used for this study comes from Hugging Face and Kaggle, which includes a comprehensive collection of Stack Overflow questions and their corresponding tags. We used preprocessing steps such as tokenization, normalization, and padding to prepare the data for training. Traditional ML models like Naive Bayes, Support Vector Machines (SVM), Random Forest, and K-Nearest Neighbors (KNN) were initially explored, yielding an accuracy of around 70%. However, a hybrid model incorporating Gated Recurrent Units (GRU), Convolutional Neural Networks (CNN), and attention mechanisms significantly outperformed these models, providing better accuracy and computational efficiency. The findings of this paper highlight the potential of integrating natural language processing (NLP) techniques to automate the tagging process on platforms like Stack Overflow, improving the organization and searchability of content.