Investigating best Deep Learning architectures for merge conflict resolution data

Context

Merge conflict resolution is a critical challenge in software development, particularly in large, collaborative projects that use version control systems like Git. When multiple developers modify the same part of a codebase, conflicts arise that require manual intervention. Existing automated resolution strategies often rely on rule-based approaches or traditional machine learning models, which struggle with complex and ambiguous cases. Deep Learning has the potential to improve conflict resolution by learning patterns from historical merge conflicts and predicting optimal resolution strategies. However, identifying the most effective Deep Learning architecture for this task remains an open question.

Motivation

The primary problem is that current automated merge conflict resolution methods are limited in accuracy and adaptability. Traditional techniques often fail to generalize across different repositories and programming languages, leading to high rates of incorrect resolutions or excessive manual interventions. Deep Learning models, such as transformers, graph neural networks, and sequence-based architectures (e.g., LSTMs, CNNs), offer promising solutions but require thorough investigation to determine which architecture performs best in terms of accuracy, explainability, and computational efficiency. This research aims to identify the most suitable Deep Learning approach for improving this kind of tabular dataset on merge conflict resolution.

Goal

The goal of this research is to evaluate and compare different Deep Learning architectures for merge conflict resolution using a dataset of historical conflicts extracted from Git repositories.

Requirements

The student should have:

Pointers