Prompter

doi:10.1007/s10664-015-9397-1

Prompter

2015-08-30

32

PRE

AI

Massimiliano Di Penta

DOI：10.1007/s10664-015-9397-1

Original

Original request for help

Save

Abstract

En 中文

Developers often require knowledge beyond the one they possess, which boils down to asking co-workers for help or consulting additional sources of information, such as Application Programming Interfaces (API) documentation, forums, and Q&A websites. However, it requires time and energy to formulate one's problem, peruse and process the results. We propose a novel approach that, given a context in the Integrated Development Environment (IDE), automatically retrieves pertinent discussions from Stack Overflow, evaluates their relevance using a multi-faceted ranking model, and, if a given confidence threshold is surpassed, notifies the developer. We have implemented our approach in Prompter, an Eclipse plug-in. Prompter was evaluated in two empirical studies. The first study was aimed at evaluatingPrompter's ranking model and involved 33 participants. The second study was conducted with 12 participants and aimed at evaluating Prompter's usefulness when supporting developers during development and maintenance tasks. Since Prompter uses volatile information crawled from the web, we also replicated Study I after one year to assess the impact of such a volatility on recommenders like Prompter. Our results indicate that (i) Prompter recommendations were positively evaluated in 74 % of the cases on average, (ii) Prompter significantly helps developers to improve the correctness of their tasks by 24 % on average, but also (iii) 78 % of the provided recommendations are volatile and can change at one year of distance. While Prompter revealed to be effective, our studies also point out issues when building recommenders based on information available on online forums.

Keywords:

Recommenders

Mining software repositories

Stack overflow

Empirical studies

Prompter

Abstract

Journal

Organization