Replication package for the paper "Multi-language Software Development in the LLM Era: Insights from Practitioners' Conversations with ChatGPT"

This is the replication package for the paper “Multi-language Software Development in the LLM Era: Insights from Practitioners’ Conversations with ChatGPT”.

This research was done using the 2023-10-12 release of DevGPT; for an overall explanation of the dataset, please consult the original work at: DevGPT/snapshot_20231012 at main · NAIST-SE/DevGPT (github.com).

After downloading the snapshot, extract the archives into the same folder of the notebooks.

The "analysis_single_source.ipynb" script collects all metrics for every file individually. The variable "fileNumber" selects which JSON to use. For overall metrics, the "analysis_all_sources.ipynb" file was used.

The normalization process is inside both notebooks and can be found below the 'Normalization' tag.

The folder "open_card_sorting" contains the 3 participants' Opencard results and analysis.

The folder "results" contains all the CSV files produced by the scripts.

The folder "normalized" contains the individual datasets already preprocessed through the scripts.