XSEDE ECSS Symposium: Exploring New Levels of Complexity in ‘Hyper Glyphs’ Reveal New Insights in Visualized Data
By Jan Zverina, San Diego Supercomputer Center/UC San Diego
|These two hyper glyph towers, based on visualizations of the Lahman Baseball Database, show the hits and home runs of Babe Ruth (left) and Barry Bonds (right) throughout their careers, from the bottom (earliest) to top (latest). Note how the yellow cubes in the Ruth visualization are more consistent throughout his career with an expected tapering toward the end, while the Bonds visualization shows a late-career surge that revealed his alleged use of performance-enhancing steroids. Images courtesy of Jeff Sale/SDSC.|
While the use of glyphs in scientific research is well documented, their untested limits as to how much information can be contained in a single glyph – or a ‘hyper glyph' – has the potential to lead to new knowledge and insights through data visualization, according to a presentation at XSEDE's first ECSS (Extended Collaborative Support Services) Symposium of 2019.
"There's some wild stuff I'll be showing you," said Jeff Sale, a visualization consultant for XSEDE's ECSS group and a research programmer/analyst with the San Diego Supercomputer Center (SDSC) at UC San Diego, an XSEDE Service Provider. "We're hoping to reach folks who are looking deep into their data sets and don't know where to begin when it comes to learning how to visualize the results of such data."
Sale's presentation covers the long evolution of the glyph, which is widely known as a typographic terminology used to describe visual symbols or markers. Over time, with advances in technology, glyphs have been used to describe many new forms, such as "gamer glyphs" which are widely used by video game players. Moreover, the use of emoticons that millions of people include in texts and emails is even more widespread, and considered to be one of the newest species of glyphs.
In the world of scientific research, even a wind barb – a symbol which represents a certain speed and direction of wind – is considered a glyph, said Sale, who also works in XSEDE's Workforce Development area and has an extensive background in data visualization.
"A growing percentage of the ‘big data' torrent consists of semi-structured, unstructured, and non-traditional data, presenting a challenge for conventional visualization methods," according to Sale. "Hyper glyphs add a new dimension to how researchers can not only find new insights in their data, but present their findings in a way that resonates with those who are inclined to better understand concepts, results, and trends if they're presented in a visual format."
The primary thrust of Sale's presentation focused on a video demonstration of ANTz, a free, open-source tool that provides a menu of various hyper glyph geometries and extensions so researchers can better identify trends in their data, by visually clustering similar glyphs or noting anomalies in glyph structures, for example. ANTz is designed for interactive exploratory visual analytics by anyone working with large amounts of non-standard or unconventional data.
While many of the visualizations in the video could also be used as some very impressive wall or screensaver art, they highlight the seemingly endless diversity of research domains that have used hyper glyphs for visualizations, from analyzing pedestrian and bike traffic patterns and density in urban areas, to animated research such as seismic wave propagation or tracking wind patterns.
From Hits and Home Runs to Bits and Bots
Sale highlighted two hyper glyph demonstrations in particular. The first was a visualization of a statistical analysis of the hits and home runs of baseball players, called the Lahman Baseball Database, used in part to analyze the ebbs and flows in their entire batting careers. "Among other things, this data quickly showed that when fields became smaller, the number of home runs became greater," said Sale.
The visualization also verified a late surge in Barry Bonds' entire career, leading Major League officials to suspect Bonds of using steroids to boost his performance.
Also shown was a visual analysis of Twitter users' daily tweets, as part of a project Sale was involved with to detect ‘bot,' or internet robot, activity. Even though Twitter limits access to the last 3,200 tweets of each user, Sale said these animations allowed researchers to visually distinguish between "bots and not-bots" based on the glyph patterns.
XSEDE's ECSS Symposium allows the 70+ ECSS staff members as well as any XSEDE users to exchange information about successful techniques and new technologies used to address challenging science problems. Two 30-minute, technically-focused talks are presented the third Tuesday of each month, followed by a brief question and answer period. The Symposium is open to everyone.
"The Symposium is an excellent opportunity to learn more about new areas of computational science and the positive impact that ECSS staff have on important research problems," said Robert Sinkovits, XSEDE's co-director of ECSS. "ECSS is available to all XSEDE PIs, regardless of field of science or size of the allocation. Support in a wide range of areas, including visualization, can be requested either when an allocations proposal is submitted or later as a supplemental request."
More information about the ECSS Symposium and the latest presentations can be found at
https://www.xsede.org/for-users/ecss/ecss-symposium. For general information about ECSS, including guidance on requesting support, see https://www.xsede.org/for-users/ecss. To learn more about ANTz and view tutorials and introductory lessons, please see https://www.edworlds.com/antz/toroids/tutorials/intro_lessons/index.html. Please send any questions or feedback to Jeff Sale at firstname.lastname@example.org.