This website requires JavaScript.

Neural Transition-based Parsing of Library Deprecations

Petr BabkinNacho NavarroSalwa AlamirSameena Shah
Dec 2022
摘要
This paper tackles the challenging problem of automating code updates to fixdeprecated API usages of open source libraries by analyzing their releasenotes. Our system employs a three-tier architecture: first, a web crawlerservice retrieves deprecation documentation from the web; then a speciallybuilt parser processes those text documents into tree-structuredrepresentations; finally, a client IDE plugin locates and fixes identifieddeprecated usages of libraries in a given codebase. The focus of this paper inparticular is the parsing component. We introduce a novel transition-basedparser in two variants: based on a classical feature engineered classifier anda neural tree encoder. To confirm the effectiveness of our method, we gatheredand labeled a set of 426 API deprecations from 7 well-known Python data sciencelibraries, and demonstrated our approach decisively outperforms a non-trivialneural machine translation baseline.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?
0
被引用
笔记
问答