DRONE: A Tool to Detect and Repair Directive Defects in Java APIs Documentation

Abstract

Application programming interfaces (APIs) documentation is the official reference of the APIs. Defects in API documentation pose serious hurdles to their comprehension and usage. In this paper, we present DRONE, a tool that can automatically detect the directive defects in APIs documents and recommend repair solutions to fix them. Particularly, DRONE focuses on four defect types related to parameter usage constraints. To achieve this, DRONE leverages techniques from static program analysis, natural language processing and logic reasoning. The implementation is based on the Eclipse-plugin architecture, which provides an integrated user interface. Extensive experiments demonstrate the efficacy of the tool.

Exploiting Natural Language Structures in Software Informal Documentation

“…Since the automated recognition of such intentions could help developers in detecting text content useful to accomplish different and specific maintenance and evolution tasks, in a previous work we proposed DECA [19], an approach leveraging language syntactical patterns to classify sentences' intent. Similar approaches have been proposed for several other purposes (e.g., apps' reviews classification [23], [34], bug reports quality assessment [11], inconsistencies detection in API documents [45], [46]). However, the main disadvantage of such approaches is that they require the manual identification of recurrent language patterns that would be exploited for automated classification.…”

“…Chaparro et al [11] defined 154 recurrent patterns to describe observed behavior (OB), expected behavior (EB) and step to reproduce (S2R) in bug descriptions with the aim of detecting the presence (or absence) of these pieces of information in such kind of artifacts (i.e., bug descriptions). Zhou et al [45], [46] employed specific linguistic patterns for automatically detecting inconsistencies between API documents and source code. Frequent grammatical patterns (i.e., dependencies in which either the governor or the dependent is a code-like term) along with structural features were also used by Petrosyan et al [36] to discover tutorial sections that explain a given API type.…”

Scout-bot: Leveraging API Community Knowledge for Exploration and Discovery of API Learning Resources

“…Correctness. This quality dimension focuses on the validity of API documentation knowledge elements, with several works addressing the issue [14,20,56,66,71,74,75,76,84,101,108,110,111,112]. Incorrect API documentation has been reported [20,66] to cause challenges to users while using and learning APIs.…”

Beyond accuracy: assessing software documentation quality

“…Defective Software Documentation. Defect detection tools have been widely investigated at the code level, but very few studies focus on defects at the document level [22]. The existing approaches in the documentation space investigate inconsistencies between code and documentation.…”

“…Wen et al [18] presented a large-scale empirical study of codecomment inconsistencies, revealing causes such as deprecation and refactoring. Zhou et al [21,22] contributed a line of work on detecting defects of API documents with techniques from program comprehension and natural language processing. They presented DRONE to automatically detect directive defects and recommend solutions to fix them.…”