EdWordle: Consistency-preserving Word Cloud Editing

Yunhai Wang 1    Xiaowei Chu 1    Chen Bao1    Lifeng Zhu 2   
Oliver Deussen3    Baoquan Chen1    Michael Sedlmair4
1Shandong University   2Southeast University   3University of Konstanz, Germany  4University of Vienna, Austria  

IEEE Transactions on Visualization and Computer Graphics (Proc. InfoVis 2017), 2018

Figure 1: Result of a case study with a professional writer who sought to visualize a BBC news feed: the left image shows the input Wordle layout; the right image shows the layout that was created using EdWordle. The writer ordered related words into semantically meaningful groups, one group per story. Each group was organized spatially together and color-coded, creating a layout that the user referred to as a storytelling cloud”.


We present EdWordle, a method for consistently editing word clouds. At its heart, EdWordle allows users to move and edit words while preserving the neighborhoods of other words. To do so, we combine a constrained rigid body simulation with a neighborhood-aware local Wordle algorithm to update the cloud and to create very compact layouts. The consistent and stable behavior of EdWordle enables users to create new forms of word clouds such as storytelling clouds in which the position of words is carefully edited. We compare our approach with state-of-the-art methods and show that we can improve user performance, user satisfaction, as well as the layout itself.


Download MP4[37M]

System: www.edwordle.net


Paper: [PDF 11.2M].
Supp: [ZIP 49.8M].



Figure 2: Overview of our method: (a) given a Wordle, (b) we use our customized rigid body dynamics to move words close to each other; (c) if a word is moved, the forces update the words accordingly; (d) empty spaces are removed by using a local version of the Wordle algorithm.



Figure 3:Comparison between Maniwordle and our method for movement and rotation. (a) Input word cloud where the word “dedicate” is moved down and rotated about a small angle; (b,c) Results generated by Maniwordle and our method, respectively.


Figure 4: (a) Realized adjacencies for various word clouds; higher is better. The dashed and continuous lines match up well, indicating that EdWordle is able to preserve neighborhoods well. (b) Mean and standard deviation of compactness for various word clouds; higher is better. EdWordle (dotted) produces substantially more compact results. (c) An example for refining a semantic word cloud (top) with EdWordle (bottom).

Figure 5: Results of our case studies, visualizing (a) an article about how much time apps eat up, (b) a speech by Obama, (c) a transcript of an interview on a psychological topic, (d) an article about solar eclipse, and (e) a speech by Martin Luther King.


The authors would like to thank Haifeng Zhang for making the evaluation. This work is supported by the grants of NSFC-Guangdong Joint Fund (U1501255), NSFC (61379091, 91630204), the National Key Research & Development Plan of China (2016YFB1001404),Shandong Provincial Natural Science Foundation (2016ZRE27617), NSF of Jiangsu Province (BK20150634), National Foreign 1000 Talent Plan (WQ201344000169), Leading Talents of Guangdong Program (00201509), the Fundamental Research Funds of Shandong University, and the FFG project 845898 (VALID).