News is commonly intended to be delivered in an objective, unbiased manner – and therefore presented plainly and formally – even though its content often affects readers emotionally. The NewsViz system aims to enhance the news reading experience by integrating 30 seconds long Flash-animations into news article web pages depicting their content and emotional aspects. NewsViz interprets football news texts automatically and creates abstract 2D visualisations. The user interface enables animators to further refine animations. Here we focus on the emotion extraction component of NewsViz ,which facilitates subtle background visualisation. The emphasis of NewsViz lies on expression, impacting on the reader’s understanding of the article and making it more memorable. NewsViz detects moods from news reports. The original text is part-of-speech tagged and adjectives and/or nouns, the word types conveying most emotional meaning, are filtered out and labelled with an emotion and intensity value. Subsequently reoccurring emotions are joined into longer lasting moods and matched with appropriate animation presets. Different linguistic analysis methods were tested on NewsViz: word-by-word, sentence based and incremental minimum threshold summarisation, to find a minimum number of occurrences of an emotion in forming a valid mood. NewsViz proved to be viable for the fixed domain of football news, grasping the overall moods and some more detailed emotions precisely. NewsViz introduces a novel approach to a universally applicable emotion scheme which offers an efficient technique to cater for the production of a large number of daily updated news stories. NewsViz fills the gap of lack of information for background or environment depiction encountered in similar applications. Further development may refine the detection of emotion shifts through summarisation with the full implementation of football and common linguistic knowledge. Future work will reveal whether NewsViz is feasible when extended to different domains.
- Eva Hanser & Paul McKevitt School of Computing and Intelligent Systems, Faculty of Computing and Engineering, University of Ulster, Magee, UK
Full Text (PDF) p. 1061-1068