Replication Package for the Paper Entitled: "Enhancing Requirements Testability Prediction with Clustering-Based Differential In-Context Learning"

This replication package contains the materials used to support the clustering-based differential in-context learning (dICL) approach for requirements testability prediction as reported in the paper “Enhancing Requirements Testability Prediction with Clustering-Based Differential In-Context Learning.” The package includes the six requirements datasets with testability labels, source code for clustering methods (K-means with elbow method and hierarchical clustering with silhouette score), GPT-4-based zero-shot prediction code, and experimental results.