Script2Graph: Auto-construct Screenplay Text World by Mining Contextual Information

Abstract

With the rapid development of artificial intelligence and the continuous emergence of innovative technologies, its application in movie scenes has gained attention. Meanwhile, the application of natural language processing techniques in movie scripts has gradually attracted the attention of researchers, making text information extraction more intelligent and convenient. However, most of the previous works for unstructured script processing ignore the theoretical guidance of computational language and lack the framework of knowledge schema. Therefore, we propose Script2Graph method based on text world theory to solve unstructured text information extraction and knowledge organization, and realize the automatic construction of script text world. Our experiments demonstrate that the method is effective in extracting text world elements and using high-quality information in downstream applications.