I found a paper, having a GitHub repository, that describes a new methodology that can be an improvement for the famous CleverCSV Python library.