A Study on Topic Visualization Techniques for Story Telling with News Articles

Abstract

With increasing amount of data that are published online, it becomes increasingly challenging to analyze and build narratives from the data, as well as producing graphics that support the narratives. Journalists express difficulties of performing such tasks during their news article production, because it often involves interacting with data experts and artists, and repeating the process multiple times. In order to alleviate this problem, it is necessary for journalists to collect their own data without the help from other experts and perform quick analysis to yield preliminary results, in order to see if the data potentially contain newsworthy information and determine if they need to be further analyzed either by themselves or by the professionals. Also, when the analysis is complete and newsworthy insights have been found, they need to effectively deliver the findings to artists who produce graphics for the story to avoid producing irrelevant graphics or misrepresenting the underlying data. In addition, since the format of news articles is often limited by the publishing platforms, they are often restricted to producing visualizations that do not involve any interactive exploration of the data. In this thesis, we present a design study of a tool that we have designed and implemented with the help from professional journalists to support the various stages in the news article production pipeline. Especially, we provide ways to collect, analyze, and visualize the information that are extracted from past news articles for storytelling using Wordle, a word cloud visualization method known for its attractiveness and aesthetic qualities [99]. We describe the interviews with the journalists about the challenges that they encounter in writing news articles such as the stress of interaction with experts or artists and the limitations that prevent them from producing interactive visualizations using the techniques developed over time in InfoVis community. Then, we describe the design process of NewsWordle, a tool that is designed to improve the overall news production. We evaluate the tool with journalists on the field with the case studies to see how NewsWordle can be utilized in their work-flow. We found that, by aligning the visualization techniques used for both analysis and creation of the visual prototypes for artists, the journalists can reduce time and effort involved in performing each task. We conclude the thesis with discussions for designing the tools that alleviate such challenges.