University of Maryland, College Park
Spring 2005
Viz4All

A survey of Internet

Visualization Tools

This page is a survey of visualization tools examined by graduate students participating in the spring 2005 Information Visualization class at the University of Maryland, College Park. Our criteria for visualizations on this website: are designed for use by the general public (not specialists); present practical, useful, and possibly entertaining information; with an interactive interface that gives users control over the display.

For each visualization tool we present a brief synopsis of the its functionality, and a screenshot. The list of tools are sortable by: visualization taxonomy, application domain, or implementation system using the links to the left.

Initial compilation by: Xu Liu
Comments by: Sandro Fouché (sandro *at* cs *dot* umd *dot* edu)



Application Domain:

Business / Stocks

SmartMoney: Map of the Market

implementation: Java applet
application domain: Business / Stocks
visualization: hierarchical

Smart Money Magazine's presentation of the top stocks of the last 52 weeks, is an excellent example of visualization put to good use in assisting understanding. This Java applet runs smoothly with several nice features, including details on demand and a search function. While this remains one of the best visualizations listed here, it is not perfect. In particular, the ability to filter stocks users are not interested in would be welcome additions to this application.

They Rule

implementation: Macromedia Flash / Shockwave
application domain: Business / Stocks
visualization: network

This network visualization is a novel representation of the often overlooked relationships between some of America's largest corporations. While the tool provides some conveniences, such as the ability to expand, relocate, or remove nodes, several limitations of the implementation make this tool more difficult to use. First, the expansion function only operates with regard to direct adjacency, thereby increasing the number of actions required to perform certain tasks. Additionally, while it is possible to relocate a node, none of the attached leaves are moved with it, this makes even simple re-organization of the display tedious. The program is further hampered, by the graph layout implemented by the program, which increases the need for manual re-organization. Never the less, this application provides a fascinating insight into corporate America.



Application Domain:

Demographics

Dynamic Choropleth Maps

implementation: Java applet
application domain: Demographics
visualization: 2-dimensional

This implementation of Dynamic Choropleth Maps allows for geographical analysis of demographic, health, and meteorological data. The tool plots the values for any chosen data variable. potentially filtered by two other variables, as colored areas on a map of the United States. While users are free to adjust the threshold criteria for any of the three variables using double-ended sliders, there is no mechanism for changing the categorical color assignments. The chosen colors while somewhat garish, and not necessarily intuitive, are color-blind safe. The ability to swap the plotted axis with either of the filtering axis would be a welcome addition, as well as the ability to plot more than data variable using distinct color scales.

Fannie Mae: DataPlace

implementation: Dynamic HTML
application domain: Demographics
visualization: 2-dimensional

Fannie Mae provides a tool for examining a limited amount of recent Census data across a region. The application, implemented using Dynamic HTML, runs nicely, and does an adequate job of plotting the provided data. Potential pitfalls of the implementation include: a small display area, a somewhat drab (but mostly color-blind safe) palette of colors, and a lack of filtering tools.

Hive Group - World Data

implementation: Java applet
application domain: Demographics
visualization: hierarchical

This treemap representation of world population data provides an excellent exploration tool for the dataset. Presented with flexible hierarchical categorization and double-ended sliders for filtering, this visualization makes it easy to restrict the display to data users find relevant. Had the designers included a mechanism for assigning a different color scale, or even inverting the existing one, the application would be even better; as implemented certain options can yield an unintuitive assignment of colors to data values.

Name Voyager

implementation: Java applet
application domain: Demographics
visualization: temporal

NameVoyager, presented by babynamewizard.com, is a visualization of first name usage over the course of the last century. Shown as a stacked line graph over the last 100 years, the display can be intimidating at first. The search function is limited to substring searching at the beginning of a name, but the dynamic query style result display is intuitive and quick, and helps with the initial information overload. One concern about this application is the constantly shifting scales used on the Y-axis; while this makes it easier to see the trends in filtered results, the changing maximum values on the axis can be misleading.

State Cancer Profiles

implementation: Java applet
application domain: Demographics
visualization: multi-dimensional

The National Cancer Institute provides a multi-dimensional view of demographic data related to cancer. Coordinating 2 variables selected from drop-down lists with mini-geographical maps by state, county, or region, the tool provides a mechanism for exploration of complex relationships within the data. In addition to visualizing discrete values, this is one of the few tools to indicate confidence intervals for selected data sets.



Application Domain:

Geographic

Google Maps

implementation: Dynamic HTML
application domain: Geographic
visualization: 2-dimensional

A basic mapping application that features extremely fast operation (zoom and pan) as well as intelligently designed location markers. Traditional vector views as well as satellite views are available. The vector view is especially nice for the well implemented text labeling of features such as roads and points of interest.



Application Domain:

Government / Politics

Fundrace: Money Map

implementation: HTML
application domain: Government / Politics
visualization: 2-dimensional

This simple, yet effective visualization makes use of statically generated maps to provide information about the financial backing of candidates in the 2004 presidential election. Unfortunately, while the application provides a set of radio buttons and a pull-down menu to select the particular map displayed, the limitations of static images prevents this tool from becoming more useful. Dynamically generating the maps would have allowed filtering and zooming of the map to isolate detailed information. Additionally, more detailed information might have been made available to users if the implementors had supported basic image maps.

Lexis/Nexis map of available sources

implementation: Java applet
application domain: Government / Politics
visualization: hierarchical

Lexis/Nexis employs hyperbolic trees to visualize the available resources on their service. Hyperbolic trees suffer from changing perspective that makes it difficult for users to locate specific nodes within the display. While the Lexis/Nexis implementation mitigates this by maintaining a strict alphabetic organization of nodes within a given branch, the layout of a hyperbolic tree can still make this confusing. The addition of search and filtering functions would have greatly increased the usability of this tool.



Application Domain:

Internet / Search

Grokker

implementation: Java applet
application domain: Internet / Search
visualization: hierarchical

A cluster map visualization, Grokker attempts to present an alternate view of Yahoo! search results. This tool provides basic filtering on Date and Yahoo! rank as well as a search box to find pages within the visualization. The application provides a nested circles (or squares) layout of search results, which is a pretty, but functionally difficult to use interface to the data. Unlike the traditional list format, the results are not shown within the majority of the visualization. While the colors used within the application are appealing, they are used in an inconsistent manner, and can be confusing since the same color can be assigned to dissimilar clusters.

Hive Group - Google results

implementation: Java applet
application domain: Internet / Search
visualization: hierarchical

The Hive Group uses their Honeycomb toolset to create a treemap visualization of Google results. Unfortunately, the visualization is ill-suited to the underlying data. While there are ample tools for filtering the results (double sliders, checkboxes, text selection, and search), Google search results lack the hierarchal structure treemaps were designed to exploit. In the end, this visualization is not clearly better than more traditional search results.

Microsoft Netscan - Usenet crosspost networks

implementation: Java applet
application domain: Internet / Search
visualization: network

Microsoft's NetScan project has produced two visualizations of Usenet posts: a treemap display, and the cross-post visualization surveyed here. The cross-post visualization is a network diagram of postings that appeared in multiple newsgroups on Usenet, presented in a node-link diagram. Nodes filtered by the provided slider drop to the bottom of the display like virtual leaves falling from a simulated tree; a nice feature of this implementation that allows users to gauge the amount of data being filtered. Like many other network visualizations, position and color seem to be haphazardly assigned to nodes, and users are given no control over these assignments.

Touch Graph: Google results

implementation: Java applet
application domain: Internet / Search
visualization: network

Like many node-link visualizations touchgraph suffers from visual confusion caused by the many nodes and interconnections between them. While the underlying data of links between websites is interesting and possibly useful information, the lack of tools within the application for managing the wealth of data limits it's utility. Without controls to filter and search among the nodes, this visualization is is less useful than it might have been.



Application Domain:

Lexical / Text

PieSpy: Shakespeare Plays

implementation: Java applet
application domain: Lexical / Text
visualization: network

Intended as a tool for visualizing social networks in Internet Relay Chat (IRC) channels, PieSpy is demonstrated here visualizing the conversations in Shakespearian works. Fundamentally an off-line tool, the website only provides screenshots and pre-canned movies of the tools output. The network visualization presented positions nodes randomly, and their occasional movement makes the animations more difficult to follow. Conveniently, the small number of characters in a given scene, keep the number of nodes and links on screen simultaneously a manageable number, and there seems to have been some attempt made at keeping the node positions somewhat stable, with some success.

TextArc: reading Hamlet

implementation: Java applet
application domain: Lexical / Text
visualization: temporal

TextArc presents a network visualization of the text of Shakespeare's Hamlet. Node-Link diagrams, commonly used to visualize networks of information, are over-taxed here in an attempt to wrap the Bard's work in a more modern package. The result is a visually interesting, yet sometimes difficult to use interface to a well known work of literature. The word search functionality (IE only), is a welcome feature, but the program would be improved with the inclusion of filtering capabilities based on names, parts of speech, or other relevant criteria. Some novel insights into Hamlet are possible with TextArc, but the visualization is sometimes difficult to use.

Visual Thesaurus

implementation: Java applet
application domain: Lexical / Text
visualization: network

This demonstration of network visualization applied to synonyms (and antonyms) in language is a somewhat entertaining way to search for terms in a thesaurus. The interface includes many of the tools found in other node-link visualizations, as well as a suite of filters and tools well suited to the application. Parts of speech are used as node coloring as well as available was filtering criteria. These features, as well as search, history, and multi-lingual capabilities make this tool quite easy to use. Common problems with network diagrams exist here as well, including the lack of meaning in positioning, and the potential for overcrowding in the display. These problems are mitigated by the relatively small number of terms available as synonyms and antonyms, and that limitation more than any other is this tools weakness.

Word Count

implementation: Macromedia Flash / Shockwave
application domain: Lexical / Text
visualization: 1-dimensional

This dynamic animation of the most common 86800 words in the English language (as represented by the British National Corpus) is an entertaining mechanism for visualizing the data. The visually distinct animation and presentation of words is novel, but this application only provides two tools for identifying words (simple search and search by rank), which elevate this tool beyond a plain list. The option to filter the data by part of speech, and more comprehensive search capabilities would enhance this visualization's utility.



Application Domain:

Movies / Music

Glass Engine

implementation: Java applet
application domain: Movies / Music
visualization: multi-dimensional

An interesting use of sliding value indicators used to explore the musical work of Philip Glass. The slider bar widgets employed in this visualization are actually quite nice; creating a sort of user interface as visualization. Each bar gives a concrete notion of its range, the number and relationship of discreet values available. But since the bars employed here seem to operate independently of one another, the resulting interface can be slightly unintuitive to use. Finding a particular work of Mr. Glass becomes a matter of randomly sliding any one of the widgets until you encounter the piece desired. Oddly this visualization does not seem to highlight particular trends or even thematic elements within the works, although the tools for such exploration seem to be there.

Hive Group - iTunes top 100 songs

implementation: Java applet
application domain: Movies / Music
visualization: hierarchical

Here again the Hive group provides a treemap visualization built using their Honeycomb product. This treemap, presenting the top 100 songs on Apple's iTunes music site, is a full featured implementation including a variety of filter controls and a search box, as well as user selectable treemap configuration. Mousing over the area representing a song, pops up details on the song chosen. The rating for this application reflects the mismatch between the hierarchical nature of a treemap and the somewhat linear nature of a top 100 list.

InXight - top 100 movies

implementation: Java applet
application domain: Movies / Music
visualization: multi-dimensional

InXight provides a variety of demonstrations for the TableLens tool for multidimensional visualization of data. This demonstration, focused on the top 100 box-office earners does a fair job of handling allowing exploration of these movies. The tool provides basic functions such as sorting on a given column and reordering columns, as well as some minor details on demand as users mouse over entries in the table. Conspicuously missing from TableLens are search and filter controls, which reduce the usability of the product somewhat.

Live Plasma

implementation: Macromedia Flash / Shockwave
application domain: Movies / Music
visualization: network

LivePlasma's visualization of movies and music allows users to explore entertainment options related to their existing favorites. While the resulting display is colorful and even somewhat informative, the visualization shares some issues with many other node-link diagrams. Positioning and coloring within the results seems arbitrary, and more importantly, the links between nodes (representing musical acts, actors, or movies) are never explained. The inability for users to filter the results, and the disorganization of the visualization make this tool harder to use than necessary. On the other hand, it does present some interesting results and is worth exploring.



Application Domain:

News

10x10

implementation: Macromedia Flash / Shockwave
application domain: News
visualization: 2-dimensional

A 2-dimensional image map of the top 100 news stories of the day, 10x10 is a novel visualization. The visual nature of the topic selection makes it a more interesting way to browse the news; unfortunately there tool lacks sufficient de-duplication facilities leading to images displayed several times. Some form of filtering would make it easier to find stories of particular interest for users; and variable sized images to indicate the relative importance or frequency of a story would be appreciated.

News Is Free: Top News

implementation: Java applet
application domain: News
visualization: hierarchical

A nicely implemented treemap of headline news culled from several sources, this visualization features several controls for filtering, including double-ended sliders and a search box. Built on the Honeycomb framework from the Hive group, NewsIsFree supports reconfiguring the treemap using pull-down menus at the top of the display. Here details on demand can sometimes be the only way to retrieve any information about stories represented by extremely small cells.

NewsMap

implementation: Macromedia Flash / Shockwave
application domain: News
visualization: hierarchical

Newsmap is a colorful treemap visualization of the Google news headline stories. Newsmap includes several features that distinguish it from NewsIsFree. The categorical color palette is well chosen as the headlines are separated into traditional newspaper sections. The ability to focus on a particular international or categorical view of the news also mirrors more traditional news sources. While the authors should be commended on attempting to include the story headline in every cell, the extremely small fonts and unusual word breaks employed to facilitate this can make the visualization difficult to read. Also, the lack of more flexible story filtering or a search facility leaves users occasionally struggling with the tool.



Application Domain:

Product Search / Shopping

Hive Group - Amazon product search

implementation: Java applet
application domain: Product Search / Shopping
visualization: hierarchical

Another treemap visualization built upon the Hive Group's Honeycomb toolset, this version maps various amazon products by price and manufacturer. The treemap provides appropriate filtering using: pull-downs, double sliders, and checkboxes; which can be useful since the display often includes many products without valid price data. While the implementation is well done, including excellent details on demand provided by mouse over, this visualization is not particularly suited to the task. Since there is no mechanism for comparing products on based on actual features, instead of just price or ratings, the visualization does not necessarily make it easier to choose a product.

InXight - 2001 Santa Clara Housing Market

implementation: Java applet
application domain: Product Search / Shopping
visualization: multi-dimensional

Another TableLens demonstration from InXight, this visualization of the 2001 housing market in Santa Clara, CA includes a relatively large data set. The tool provides basic functions such as sorting on a column and reordering columns, as well as some minor details on demand as users mouse over entries in the table. While TableLens is able to maintain its speed and functionality with a much larger dataset than in the other demos, the lack of search and filtering tools in this application limits it's utility.

Peet's Coffee: Coffee Selector

implementation: Java applet
application domain: Product Search / Shopping
visualization: hierarchical

A treemap visualization, here put to work in service of coffee selection. While the tool lacks filtering, the hierarchical configuration is user-selectable thought a pair of drop-down selections. While this isn't clearly the best way to visualize coffee, the colors matched to the final result of brewing the beans is a nice touch



Application Domain:

Social Networks

CiteSpace

implementation: Java applet
application domain: Social Networks
visualization: network

Intended to highlight patterns in scientific literature, CiteSpace suffers from many of the same issues that affect most network visualizations. Sporting dynamic layout and copious user adjustment facilities, the networks occasionally oscillate endlessly or collapse onto themselves. The meaning of the position and relationships between nodes are not made clear to the user, and the network can devolve into a tangle of interconnected lines. While layout issues hamper this program when working with small amounts of data, the lack of adequate filtering makes this tool difficult to use on large or even medium-sized data sets.

NetVis

implementation: Java applet
application domain: Social Networks
visualization: network

a simple node-link visualizing tool, NetVis is intended to help users map social networks. Unlike many other tools in this survey, NetVis only includes simulated data for visualization, but users are welcome to input their own dataset for modeling. Netvis is a relatively straight-forward implementation of node-link diagramming, without search or filtering capabilities which would have improved its usability.



Application Domain:

Sports

InXight - SuperBowl predictor

implementation: Java applet
application domain: Sports
visualization: multi-dimensional

Provided as a demonstration of the multi-dimensional capabilities of their TableLens product, this tool provides a tabular visualization of NFL football statistics. Here the application's ability to sort on columns allows for some quick analysis of the data, but like the other TableLens demos, there is a lack of filtering and search tools. Additionally, this particular visualization did not allow for resizing which made the tool more difficult to use.