Convolutional Neural Networks over Tree Structures for Programming Language Processing

TL;DR

This paper introduces a novel tree-based convolutional neural network (TBCNN) designed to effectively process programming languages by capturing their structural information through abstract syntax trees, outperforming existing models.

Contribution

The paper presents a new TBCNN architecture that applies convolution over program syntax trees, specifically tailored for programming language analysis tasks.

Findings

TBCNN outperforms baseline neural models in program classification.

TBCNN effectively detects specific code patterns.

TBCNN demonstrates versatility across different program analysis tasks.

Abstract

Programming language processing (similar to natural language processing) is a hot research topic in the field of software engineering; it has also aroused growing interest in the artificial intelligence community. However, different from a natural language sentence, a program contains rich, explicit, and complicated structural information. Hence, traditional NLP models may be inappropriate for programs. In this paper, we propose a novel tree-based convolutional neural network (TBCNN) for programming language processing, in which a convolution kernel is designed over programs' abstract syntax trees to capture structural information. TBCNN is a generic architecture for programming language processing; our experiments show its effectiveness in two different program analysis tasks: classifying programs according to functionality, and detecting code snippets of certain patterns. TBCNN outperforms baseline methods, including several neural models for NLP.