Pointing to Subwords for Generating Function Names in Source Code

TL;DR

This paper introduces two strategies for copying low-frequency subwords to improve automatic function name generation from source code, significantly enhancing performance on Java datasets.

Contribution

It presents novel copying strategies for low-frequency subwords, addressing out-of-vocabulary issues in function name generation models.

Findings

Improved F1 and accuracy scores on Java-small and Java-large datasets.

Effective handling of low-frequency and out-of-vocabulary subwords.

Enhanced model performance over conventional methods.

Abstract

We tackle the task of automatically generating a function name from source code. Existing generators face difficulties in generating low-frequency or out-of-vocabulary subwords. In this paper, we propose two strategies for copying low-frequency or out-of-vocabulary subwords in inputs. Our best performing model showed an improvement over the conventional method in terms of our modified F1 and accuracy on the Java-small and Java-large datasets.