Abstract
ContextDue to the unique characteristics of comics as a visual storytelling medium that embraces a wide range of genres and aesthetic styles; comic recognition (CR) presents difficulties. By utilizing topic modeling with Latent Dirichlet Allocation (LDA), this study seeks to explore the underlying themes and topics associated with CR. The dataset used for the study consists of 490 articles that were published between the years 2004 to 2023. ObjectiveTo locate and link (semantic map) the primary research fields, core research zones, and research trends that are driving the CR domain. Identifying well-known recognition models based on the collected study topics. To create the development chart from the study patterns those were collected to guide future activities in this area. MethodAn exhaustive exploration was carried out within esteemed digital libraries to curate a corpus. This meticulously collected corpus was subsequently harnessed to facilitate topic modeling. The technique of LDA was employed for topic modeling, aligning with the specific objectives set out for this endeavor. ResultThis study demonstrated a concise extraction of topics by effectively employing LDA to identify underlying trends in CR. The study also shed light on the dynamic growth of these topics through time and revealed connections across various modalities within the topic. The coherence scores for the two, five, and ten-topic solutions were discovered to be 0.39, 0.49, and 0.41, respectively. Emphasis was placed on the practical implications of LDA-based topic extraction, along with insightful observations that suggest potential directions for future research endeavors.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call