Analysis of Question-Posting Websites for Building Dialogue Models Specific to Programming Contexts

作者：Tadashi Ohara, Tomonori Hashiyama · 发表于：2025 6th International Conference on Big Data Analytics and Practices (IBDAP) · 年份：2025 · DOI：10.1109/IBDAP65587.2025.11145742

This study aims to develop a dialogue model designed for implementation in a local environment independent of major AI systems. It is in response to the increasing demand for support at programming education platforms, driven by MEXT's introduction of compulsory programming education. This study collected 48,083 Python-related questions and answers from the Japanese programming Q&A site, teratail. We then applied methods such as morphological analysis, TF-IDF, clustering, and LDA topic modeling to analyze the common challenges faced by questioners, their skill levels, and the distribution of question categories. Our analysis revealed that keywords like 'Error' and 'Method' were frequently used, and the questions' content suggested a focus on both beginner and advanced levels of programming. These findings provide key insights for building localized, high-quality response generation techniques.