On the Importance of Building High-quality Training Datasets for Neural Code Search

TL;DR

This paper emphasizes the critical role of high-quality datasets in neural code search, revealing noise issues in existing datasets and proposing a novel semantic data cleaning framework that significantly enhances model performance.

Contribution

It introduces the first semantic query cleaning framework for code search datasets, improving data quality and neural model effectiveness.

Findings

Filtering improves DeepCS model performance by 19.2% MRR

Semantic cleaning reduces query noise significantly

Enhanced datasets lead to better real-world code search results

Abstract

The performance of neural code search is significantly influenced by the quality of the training data from which the neural models are derived. A large corpus of high-quality query and code pairs is demanded to establish a precise mapping from the natural language to the programming language. Due to the limited availability, most widely-used code search datasets are established with compromise, such as using code comments as a replacement of queries. Our empirical study on a famous code search dataset reveals that over one-third of its queries contain noises that make them deviate from natural user queries. Models trained through noisy data are faced with severe performance degradation when applied in real-world scenarios. To improve the dataset quality and make the queries of its samples semantically identical to real user queries is critical for the practical usability of neural code search. In this paper, we propose a data cleaning framework consisting of two subsequent filters: a rule-based syntactic filter and a model-based semantic filter. This is the first framework that applies semantic query cleaning to code search datasets. Experimentally, we evaluated the effectiveness of our framework on two widely-used code search models and three manually-annotated code retrieval benchmarks. Training the popular DeepCS model with the filtered dataset from our framework improves its performance by 19.2% MRR and 21.3% Answer@1, on average with the three validation benchmarks.