From Cluster Hypothesis to Semi-Supervised Clustering: Exploiting LLMs to Solve the Next Release Problem

A: Datasets

            A1. Zoom

            A2. WebEx

            A3. Microsoft 365

            A4. Discord

A dataset file contains multiple NRP (next release problem) instances: each instance per sheet. Regarding the labels, “1” means the correct NRP solutions, “0” means the candidate requirements that are not part of the next release, and “-1” means the already implemented requirements.

B: Python Code

            B1. Cluster_hypothesis.ipynb

            B2. LLM.ipynb

            B3. PCKMeans.ipynb

C: Cluster vs LLM Graphs

            C1. Average F-measure results of clustering and LLM