Methods and techniques for data collection in automatic scene script composition

An exploratory study was conducted to address the problem of developing a precise system capable of generating computational narration for short scenes that may be turned into short videos using text. Literature suggests that there is no singular data source available for this purpose; therefore, a comprehensive analysis of requirements is necessary. This study analysed sources of data collection, storage, retrieval, query and indexing techniques for constructing computational narrative engines. Aside from data-related considerations, we investigated the system's architecture. A system should support a variety of user interactions, such as voice commands and text input. To evaluate the quality and effectiveness of the computational narration generated by the system, evaluation metrics should be established. The case study presented and experimental results demonstrate the effectiveness of the proposed methods. The collected dataset is used for automatic scene script composition models, enabling the generation of coherent and contextually relevant scene scripts.