Roshan Koirala is a doctorate student in physics at University of Alabama. His doctoral research focuses on modeling the quark gluon plasma using holographic principle. Besides that, Roshan is also interested in machine learning and statistical modeling.
Computational Essay: < Visualization of harmonic oscillator solution >
Project: Dynamical word cloud
The word cloud is a popular text data visualization tool showing objects sized according to there weights. However, when several representations are made with different weights, the positions of the object can change drastically. Stabilizing the position of the words in a word cloud for dynamical data is a challenging issue. This project aims to solve this problem.
Summary of Results
A general algorithm for a static word cloud is to fill the space with objects from larger to smaller size, starting from the center and spiraling outward until there is no more overlapping. For two consecutive frames, minimizing the displacement of each word between consecutive time lapse is achieved by taking the updated word data size order, placing the largest word at the center of the graph and then starting with the next largest word from the center and spiraling outward until there is no overlap with previously placed words. This process is repeated with each progressively smaller word. If the current word was present in the previous word cloud, start spiraling from its previous position until there is no overlap with already placed words. If it was not present, start from the center of the graph.
I plan to do detail tuning of the hyper-parameter for better performance, enforce stronger packing in instances where there is too much white space, add a step that pulls the words toward the center until they overlap, and assign the weight such that the word will get penalized by placing it away from the center of the graph.