10 x 10 - UC Berkeley School of Information



Information Visualization Critique of 10x10



About the Visualization

This visualization allows the user to see the top 100 most frequent news items in a one-hour increment of time by displaying the 100 top words and photos associated with world news headlines. The site intends to capture a moment in time that allows users to see what was important at that moment.

The target audience is anyone interested in the visualization of term frequency in world news. (Note, data is pulled from three sources: BBC, NYT, Reuters)

[pic]

10x10, an interactive exploration of the words and pictures that define the time.

Analysis

Visualization Structure

The visualization contains a 10x10 grid of thumbnail images. The grid does not have a visible axis, but the images are organized such that the first is in the upper left corner and 100th is in the bottom right corner. Along the right edge of the grid is a list of words from one (top) to 100 (bottom). There are two data elements per rank – an image and a word. The data elements are ordinal by ranking relationship. One hour of data is visible at a time unless searching the archives.

This visualization demonstrates a location probe view transformation as described by Card, Mackinlay, and Schneiderman. In this visualization, the location probe allows the user to investigate additional data by clicking on an image or its related word for associated news headlines. Clicking either the image or the word brings up a detail pane on the photo grid. The detail pane expands the thumbnail and lists headlines containing the associated word.

[pic]

The detail pane displays world headlines associated with the image and word.

Visual Mappings

Size & Color - Large text size and red color determine the word in focus.

Zoom – words zoom from illegible to very legible. Words above and below the one in focus decrease in size the farther away it is from the one in focus.

Orientation – A vertical list is consistent with our notion of the highest rank on top.

Color - Images have a red border and a light screen when in focus.

Frequency Pattern

This visualization allows the user to see how frequently the same image repeats on the photo grid. Detecting the repetition is enhanced by keeping thumbnail images the same dimensions and distances from each other. Uniform image size requires less effort to detect repetition of the images. Perceptually, the frequency of images gives additional information – how important or frequent a news item is in the world at a specific hour. Data updates every hour so the patterns change over time. This effect is lost when time span increases from one hour to one day, month, or year.

Critique

The concept for this visualization is extremely interesting and compelling. In the developer’s own words, “Over the course of days, months, and years, 10x10 leaves a trail of these hourly statements which, stitched together side by side, form a continuous patchwork tapestry of human life.” Unfortunately, this statement is only partially supported with this visualization.

Areas for Improvement

• There is neither a clear title on the page nor labels on the axes. This would easily orient the user to the purpose of the site.

• The visualization does not offer any mechanism for side-by-side comparisons across time increments. For example, I cannot compare the current hour with the same hour yesterday to see relationships or trends over time.

• Images are organized in rank order from left to right. Words are organized vertically from top to bottom. These different orientations make it difficult to relate the data elements, not to mention the thumbnail rank is not obvious until the user interacts with the words, activating the focus ring.

• Speaking of focus, focus occurs simultaneously in two different locations with movement along opposing axis. The focus ring around the thumbnail images moves horizontally across rows while focus moves vertically down the word list. This is distracting and disorienting. Two problems may be solved by reorienting the image rank vertically from top to bottom creating a vertical orientation for both data elements.

• This idea could be taken one step further by relocating the two data elements such that the term and the picture are closely associated, either above or below the image. Rolling over the image/term combination could slightly increase the size of both to designate emphasis and to allow the reader a better view of what is in the picture. The thumbnail image size is inadequate for distinguishing the contents of detailed images, panoramic scenes and the like although the size conserves space.

• The illegible words are a problem. The user must constantly mouse over the list to see what terms are on it. As short-term memory can accommodate no more than seven or eight chunks information, keeping a list of 100 terms in mind is an impossible task. As space is an issue, the list could be redesigned to zoom many more than just the 10 legible terms. Reducing the size of the term in focus might allow for increased size elsewhere.

Aspects that Work Well

• The visualization has a clean design that is not too cluttered with information. The user can access detailed information for further investigation with the location probe view transformation. This allows the user to see exactly what headlines feed that particular term and image.

• Additional data is available through an archive. Visualizations for previous years, months and days are available.

• The focus color and size change for the word list works well and the order on the list. This zoom effect correspondingly brings an image in focus. With some effort, the association between the two is made.

• The day, date, and time label is clear and lets the user know the time increment is Eastern Standard Time.

[pic]

10x10 archives, bottom of the screen, allow the user to search by hour, day, month, or year.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download