一句话：提出 AutoPyDep，用图网络与深度学习为 Python 依赖管理提供发布类别与日期预测推荐。
现代软件开发中依赖管理日益复杂，常导致版本冲突、构建失败与运行时错误等“依赖地狱”问题。AutoPyDep 是一个面向 Python 库依赖管理的推荐系统，提供依赖分析、关系映射以及发布类别与日期预测。作者将 23 个 Python 库的发布说明转换为图网络，结合 NLP 技术与基于社区的 deepWalk 算法生成嵌入，用于发布类别预测和发布日期预测。系统采用投票分类器取得 0.8 的 F1 值，ARIMA 模型的平均绝对误差为 1.8 个月。该工具旨在提升依赖管理准确性，为开发者提供可操作的洞察，支持软件开发中的决策。
Managing software dependencies is increasingly challenging due to the complexity of modern development, often resulting in "dependency hell" with version conflicts, build failures, and runtime errors. To address these issues, we present AutoPyDep, a recommendation system for Python library dependency management. AutoPyDep features dependency analysis, relationship mapping, and predictive modeling for release categories and dates. By transforming release notes from 23 Python libraries into a graph network, we leverage NLP techniques and a community-based deepWalk algorithm to generate embeddings for tasks such as release category prediction and release date forecasting. Key contributions include a voting classifier achieving a robust F1 score of 0.8 and an ARIMA model with a Mean Absolute Error (MAE) of 1.8 months. AutoPyDep enhances dependency management accuracy, offering actionable insights for developers and supporting improved decision-making in software development. A demonstration of our tool is shared in this link.
@inproceedings{paperbot9159,
title = {AutoPyDep: A Recommendation System for Python Dependency Management Utilizing Graph-Based Analytics},
author = {Dibyendu Brinto Bose and Travis Chan and Matthew Trimble and Chris Brown},
booktitle = {FSE},
year = {2025},
doi = {10.1145/3696630.3728576}
}