Applying Topic Modelling to the Folk Song Corpus of the Podillia Region

This dataset is based on Ukrainian folk songs from the Podillia region collections (Dei 1965; Yefremova & Dmytrenko 2014; Myshanych 1976). The text corpus comprises 2,762 songs, containing a total of 52,004 lines and 209,075 tokens.

The corpus of Podillia folk songs is in Ukrainian. It was analysed using the R programming language (version 4.4.1) along with RStudio (version: 2024.04.2+764).

The text analysis code was developed at the Estonian Literary Museum.

This dataset includes the following files:

1. TM_Ukr.folk_songs.R:  R script for the topic modeling analysis of the corpus. It covers text preprocessing, stopwords removal, tokenization, lemmatization, part-of-speech (PoS) tagging, document-term matrix (DTM) construction, latent Dirichlet allocation (LDA) analysis, coherence evaluation metrics, word embeddings using the GloVe algorithm, and K-means clustering.

2. corpus_Podillia_folk_songs.csv: A file containing the text data of Podillia region folk songs.