Loom

 

Loom is a visualization tool for Usenet newsgroups. It represents the pattern and textures of events in a threaded newsgroup as a weave of a digital fabric.

 

The objective of this project is to observe patterns in key events of newsgroups. Examples of such events are: entry and exit of participants in conversation, the birth and death of subject threads, the tone of messages, and the path traversed by users as they create this social fabric.

 

Loom began as a class project. The first implementation of Loom focused on visualizing the mood or tone of messages. Current work includes creating a reconfigurable interface for the mapping of the usenet data onto the canvas.

In the following case, the left axis represents a different user as they've participated in the newsgroup. The bottom axis represents time from left to right. Lines connect messages that belong to the same thread. The user can move through time by moving the viewing window left or right. The colors represent different subjects. They can also be set to denote different suffixes (ex. .com, .edu, .org, .net, etc.) within the population of the group. (The following image was created from the newsgroup, comp.lang.java.gui.)

Figure 1

 

Different representations are displayed at different resolutions of the information. For example, the user is first presented with an image representing the entire conversation for the designated start and end time. (The following figures are from the newsgroup, soc.culture.greek.)

Figure 2

 

This is essentially an overview of the conversation. Areas of higher intensity imply more messages at that time. Rows of high intensity signify users that make many contributions to the list.

The user may also sift through the posters and order them. In the following image, the frequent posters have been sifted towards the top.

Figure 3

 

From this point, the user can click to zoom into a region of interest. The chosen point becomes the center of the new window. The resulting image is shown below.

Figure 4

 

Comparing this image to that of the java newsgroup in Figure 1, we can see that this one is more intricate in its threading. The java newsgroup contained primarilly short paths in its threads. That is, a question and an answer. In this newsgroup, the threads move rapidly to different users. There also appear to be several stand alone messages. A user slightly below the middle of the image in Figure 2 appears to post many messages that do not become incorporated into threads. The threads can be selected and followed using the keyboard (see Figure 5).

Figure 5

 

Another mode (Figure 6) allows the user to view which messages are replies and which are initial posts. In this visualization, the red units indicate replies; the white indicate initial posts. From this we can see that some posters tend to reply within existing threads far more than they initiate them, while others primarily start threads.

Figure 6

 

In a different representation, we simply see the messages with lines separating individual days. In this case, the user can modify the timescale moving these vertical lines.

 

Figure 7

 

At the lowest level in the Loom canvas, the user can access the text to the individual postings.

 

Figure 8

 

Discussion

This above implementation preserved the relationship between a thread in a usenet converstation and a thread in a fabric. Continuing work in this visualization is focusing on :

  • using the thread metaphor of the fabric to represent the path of a person within the group.
  • grouping people spartially that speak amongst each other frequently.
  • allowing for the switching of paramers in the display( newsgroups, mode of display: tone, timescale, etc) in an intuitive manner.
  • abstractions of the representation of the newsgroups, such as the image at the top of this page.

 

This visualization provides a metaphor for understanding the contents of a newsgroup of different types. It also provides a means for traversing the threads and discovering the individual postings. By representing the postings in this form, it is capturing not only the patterns of usage, but also a historical context for the postings. Combined, the different threads tell a story. It would be interesting to hold a tapestry representing '1998' and compare it to previous years.

 

 

 

Karrie Karahalios
Advisor: Professor Judith Donath
Undergraduate Researcher: Todd Kamin

Sociable Media Group