Narrative based Topic Visualization for Chronological Data

Abstract

This paper proposes several methods visualizing topics in documents, (i) Word Colony represents the dependency relationships among term occurrences in a target document. It helps users get an overview of a document. Using Word Colony with pictures gives users a more intuitive impression, (ii) topic sequence is a concatenation of Word Colonies for segmented documents. It shows plots as a story's topic transitions. (Hi) topic matrix represents relations among topics based on latent contexts within a collection of documents. Several visualization techniques enable users to visualize topics in a document in different ways. It gives variations of viewpoint and triggers the creative imagination.