This university project explores authorship identification in Romanian poetry using Natural Language Processing (NLP) techniques. The goal was to determine the author of a poem by analyzing writing style and word usage patterns.
Two main approaches were implemented:
TF-IDF + Cosine Similarity — achieved ~90% accuracy for longer poems
Bag-of-Words + Multinomial Naive Bayes — achieved ~80% accuracy
The dataset was automatically extracted from Romanian Voice using web scraping tools, and the models were trained and tested on this data.