Artificial Intelligence for Sentiment Analysis of Persian Poetry
A new AI study uses GPT-4o and BERT models to quantify the emotional sentiment in centuries-old Persian poetry.
A research team from multiple institutions, including authors Arash Zargar and Farzad Khalvati, has published a novel study applying modern large language models (LLMs) to the nuanced task of Persian poetry analysis. Their paper, "Artificial Intelligence for Sentiment Analysis of Persian Poetry," tested multiple models, including Bidirectional Encoder Representations from Transformers (BERT) and Generative Pre-trained Transformer (GPT) architectures, specifically finding OpenAI's GPT-4o to be a reliable tool for this complex linguistic task. The core objective was to evaluate if AI could grasp the intricacies of classical Persian verse and explore potential links between a poem's emotional tone and its structural meter.
The research focused on the works of two literary giants: the 13th-century mystic Jalal al-Din Muhammad Rumi and the 20th-century poet Parvin E'tesami. The AI-driven sentiment analysis yielded quantifiable results, revealing that Rumi's poems generally express happier sentiments compared to E'tesami's. Furthermore, the analysis of poetic meters indicated Rumi's superior use of rhythmic structure to express a broader spectrum of emotions. This study is significant as it moves beyond simple text generation, demonstrating that LLMs like GPT-4o can be effectively applied to advanced computer-based semantic studies in humanities research, potentially reducing human interpretive bias in literary analysis.
- The study used multiple AI models, including BERT and GPT variants, with GPT-4o identified as reliably capable of analyzing complex Persian poetry.
- Quantitative sentiment analysis revealed Rumi's poems express happier sentiments than those of Parvin E'tesami, and Rumi used meters to convey a wider emotional range.
- The findings validate LLMs for objective, computer-based semantic studies in the humanities, offering a tool to reduce human bias in literary interpretation.
Why It Matters
It demonstrates AI's potential as a powerful, unbiased research tool for quantitative analysis in the humanities and literary studies.