LYRICEL: A Rule-Augmented Artificial Intelligence-Empowered Cultural E-Learning with GPT and Machine Learning for Song Lyrics
Dimitrios P. Panagoulias, Dionisios N. Sotiropoulos, Konstantina Chrysafiadi, Evangelia-Aikaterini Tsichrintzi, Evangelos Sakkopoulos, George A. Tsihrintzis, Maria Virvou
Abstract
In this paper, we present LYRICEL, an advanced AI-enhanced system that combines a rule-based decision-making mechanism, OpenAI's application programming interface (API), and additional external machine learning and analytical APIs to deliver song lyrics recommendations for an e-learning platform dedicated to Greek music and songs. This new component enhances MUSILYAN, a specialised software tool designed for musicological and lyrical analysis, by complementing it with GPT4o's capabilities. While ChatGPT excels in natural language understanding and generation, MUSILYAN provides structured semantic content management and thematic organization of lyrics. The integration of the two modules in LYRICEL, holds significant promise for enhancing the exploration and understanding of poetry and lyrics, offering a robust framework for enriching cultural heritage e-learning experiences.