• Abdusamiev Dilmurod Abdugani ugli

  Author

The rise of digital humanities and artificial intelligence has opened up new lines of inquiry in literary research, giving scholars access to the analytical tools that allow processing of extensive textual corpora in ways that classical close reading could not. This paper investigates the use of AI in digital humanities inquiry, such as neural modeling, stylometry, detecting sentiment, and topic modeling. It suggests that at the heart of contemporary literary scholarship lies a use of computational methods that does not negate interpretive judgment, nor position computation as anti-criticism. Instead, this article utilizes computational inquiry as one of the steps in the analytical chain that incorporates text data as input and makes claims about the analyzed corpus as output. The article proposes a four-layer approach that connects the process of archiving texts, computational preprocessing, AI-based interpretational models, and generating scholarly products, combined with the feedback loop that allows the use of literary insights in future modeling efforts. The paper also contrasts five types of approaches in analytical literature: traditional close reading, rule-based processing, statistical stylometry, standard machine learning, and deep neural approaches in digital literary studies. The study assesses approaches along the three parameters, including analytical accuracy, scalability, and interpretative depth. The results prove neural methods do much better than tools used before; however, interpretative depth should be sought when models are accompanied by human reasoning but not treated as something already complete on own. In addition, the article situates findings in the context of the discussions around issues related to algorithmic bias and ethical issues of working with literary data, advocating for the development of research methodology that combines traditional approaches with those involving computer-assisted analysis.

• PDF