Taming Multi-Output Recommenders for Software Engineering

TL;DR

This paper proposes a new approach to improve software engineering recommender systems by emphasizing diversity, better communication, and interactive navigation to enhance developer support and decision-making.

Contribution

It introduces a research agenda for re-imagining how recommender systems communicate insights, focusing on diversity, interaction, and holistic evaluation in software engineering.

Findings

Proposes diverse and non-redundant recommendations

Highlights the importance of interactive and seamless navigation

Aims for comprehensive end-to-end evaluation methods

Abstract

Recommender systems are a valuable tool for software engineers. For example, they can provide developers with a ranked list of files likely to contain a bug, or multiple auto-complete suggestions for a given method stub. However, the way these recommender systems interact with developers is often rudimentary -- a long list of recommendations only ranked by the model's confidence. In this vision paper, we lay out our research agenda for re-imagining how recommender systems for software engineering communicate their insights to developers. When issuing recommendations, our aim is to recommend diverse rather than redundant solutions and present them in ways that highlight their differences. We also want to allow for seamless and interactive navigation of suggestions while striving for holistic end-to-end evaluations. By doing so, we believe that recommender systems can play an even more important role in helping developers write better software.