Applications of Large Language Models for the Analysis and Visualization of Story Texts
Open Access DepositedThis research explores the ability of applying Large Language Models (LLMs) to extract and analyze information from story texts. We developed an LLM-based analysis system focused on processing story texts, enabling the extraction of key elements such as characters, events, themes, symbols and motifs. With the extracted characters list, the system could further perform an in-depth analysis of character relationships within the identified character list. The analysis includes categorizing the relationship between characters (e.g., familial relationships) and assessing their affinity and strength of their connections through sentiment analysis of interactions. With the extracted event list, the system could inference causalities between events based on contextual information and prior knowledge. To enhance usability, the system visualizes the extracted information into clear and comprehensible formats such as character networks, event timelines, and other diagrams. These visualizations ensure that critical information is preserved in a clear and easily accessible manner, facilitating further analysis and interpretation by users. To evaluate the potential of LLMs in story text analysis, we employed various evaluation metrics and methods to assess the performance of several LLMs on the tasks. Through analyzing their performance and making comparisons, we derived some key insights, including the strengths and limitations of LLMs. iii Overall, this research highlights the significant potential of LLMs to automate complex story text analysis tasks. At the same time, it acknowledges the advantages and challenges of applying LLMs to literary texts, providing some insights and possible directions for future refinement and broader application.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.