Sentiment Relation Detection between Main Characters in Turkish Stories

Abstract

The rapid advancements in Natural Language Processing have facilitated tasks such as understanding texts and extracting information from them. Identifying main characters in texts or stories and detecting emotional relationships between characters is one of the current topics in Natural Language Processing. In this study, an application has been developed to identify main characters in Turkish stories and predict the emotional relationship between them using named entity recognition, dependency parsing, and sentiment analysis models. The success rate in extracting the main characters is about 77 and the success rate in determining the emotional relationship between them is about 41.