Machine Learning-Based Test Smell Detection

TL;DR

This paper proposes a machine learning approach for detecting test smells, aiming to improve over existing heuristic-based methods by leveraging a large, manually-validated dataset and evaluating multiple classifiers.

Contribution

It introduces a novel machine learning-based method for test smell detection and plans to create the largest validated dataset for training and evaluation.

Findings

Development of a large, validated test smell dataset

Evaluation of six machine learning classifiers for detection accuracy

Comparison showing potential improvements over heuristic methods

Abstract

Context: Test smells are symptoms of sub-optimal design choices adopted when developing test cases. Previous studies have proved their harmfulness for test code maintainability and effectiveness. Therefore, researchers have been proposing automated, heuristic-based techniques to detect them. However, the performance of such detectors is still limited and dependent on thresholds to be tuned. Objective: We propose the design and experimentation of a novel test smell detection approach based on machine learning to detect four test smells. Method: We plan to develop the largest dataset of manually-validated test smells. This dataset will be leveraged to train six machine learners and assess their capabilities in within- and cross-project scenarios. Finally, we plan to compare our approach with state-of-the-art heuristic-based techniques.