Compositionality-Aware Graph2Seq Learning

TL;DR

This paper introduces a compositionality-aware graph2seq learning approach using multi-level attention pooling, significantly improving code summarization performance with fewer parameters.

Contribution

It applies a multi-level attention pooling architecture to enhance graph2seq learning, demonstrating superior performance in source code summarization tasks.

Findings

Outperforms previous state-of-the-art models in code summarization.

Uses over seven times fewer parameters than existing models.

Shows that compositionality-aware architectures improve interpretability and efficiency.

Abstract

Graphs are a highly expressive data structure, but it is often difficult for humans to find patterns from a complex graph. Hence, generating human-interpretable sequences from graphs have gained interest, called graph2seq learning. It is expected that the compositionality in a graph can be associated to the compositionality in the output sequence in many graph2seq tasks. Therefore, applying compositionality-aware GNN architecture would improve the model performance. In this study, we adopt the multi-level attention pooling (MLAP) architecture, that can aggregate graph representations from multiple levels of information localities. As a real-world example, we take up the extreme source code summarization task, where a model estimate the name of a program function from its source code. We demonstrate that the model having the MLAP architecture outperform the previous state-of-the-art model with more than seven times fewer parameters than it.