Code to Comment Translation: A Comparative Study on Model Effectiveness & Errors

TL;DR

This paper compares recent source code summarization models using both automatic metrics and manual error analysis to understand their strengths and weaknesses in translating code to natural language descriptions.

Contribution

It introduces a combined quantitative and qualitative evaluation approach, including an error taxonomy, to better assess model effectiveness beyond standard metrics.

Findings

Automatic metrics do not fully capture model errors.

Manual error analysis reveals common failure modes.

Error taxonomy can guide future model improvements.

Abstract

Automated source code summarization is a popular software engineering research topic wherein machine translation models are employed to "translate" code snippets into relevant natural language descriptions. Most evaluations of such models are conducted using automatic reference-based metrics. However, given the relatively large semantic gap between programming languages and natural language, we argue that this line of research would benefit from a qualitative investigation into the various error modes of current state-of-the-art models. Therefore, in this work, we perform both a quantitative and qualitative comparison of three recently proposed source code summarization models. In our quantitative evaluation, we compare the models based on the smoothed BLEU-4, METEOR, and ROUGE-L machine translation metrics, and in our qualitative evaluation, we perform a manual open-coding of the most common errors committed by the models when compared to ground truth captions. Our investigation reveals new insights into the relationship between metric-based performance and model prediction errors grounded in an empirically derived error taxonomy that can be used to drive future research efforts