tex2net: A Package for Storytelling using Network Models

As the volume of textual data grows at a fast pace, there is an increasing need for effective techniques to analyze and present this data meaningfully. Traditional methods of summarizing text data, such as word clouds or tag clouds may not provide a comprehensive narrative overview. In contrast, visual representations, such as graphs, arguably allow the visualization of more complex information. In this paper, we propose a text-to-graph conversion technique that allows the visualization of a story's main characters and relationships. Although visualizing text data through graphs is becoming increasingly popular, existing graph tools generally depend on structured data representations and are unable to comprehensively visualize a narrative and its entities (characters). Our proposed text-to-graph conversion technique addresses this gap, by providing a valuable tool for storytelling visualization, along with relevant guidelines. To this end, we propose a methodology to learn expressive graphs (stories) by extracting relevant relationships between focal entities (characters) from a text document. Graph representation is subsequently refined to communicate the flow of sample narratives. The methodology is provided as a software library, termed tex2net. The acquired results indicate that the proposed approach is able to summarize the story, complementing the use of traditional text summarization techniques. Additionally, we found the graphical summaries more engaging and easier to understand.