Georgia Institute of Technology – Directed Exploration

VAST 2006 Contest Submission
July 2006

Authors and Affiliations:

Summer Adams, Georgia Institute of Technology, summer@cc.gatech.edu
Kanupriya Singhal, Georgia Institute of Technology, ksinghal@cc.gatech.edu

Tool(s):

We created our own tool for this contest and integrated some of the features of prefuse. Initially it was developed for a group project at Georgia Tech for an Information Visualization course in Spring 2006. Over the past couple of months, we have enhanced the functionality and further developed our theory. Additional developers of the tool for the class project that did not participate in the submission are Susan Gov and Sheena Lewis, both enrolled in graduate school at Georgia Tech. The intent of the tool is to provide an overview of the data and guide the user in their quest for connections in the data. Under that premise, we included as much of the data in the tool as possible with complementary views based on user selected filters.

 

TOC:  Who – What – Where – Debriefing - Process

VIDEO

2 page summary (for publication)     

1. WHO: who are the players engaging in deceptive activities in the plot(s)?

Name	Most relevant source files (5 MAX)
Dr. Delwin Sanderson	Scientific Review 3, Scientific Review 4, 1101163356450, picture081, picture082
Dr. Alejandro VonRyker	Scientific Review 3, Scientific Review 4, 1101163356450, picture081, picture082
Dr. Philip Boynton	1101163356450, picture081, picture082, picture083
Rex Luthor	1101163356450, picture081, picture082
Laurel Sulfate	picture083, picture084, picture085

(Note: ignore the file extensions)

2. WHAT: what is the time frame in which the plot(s) unfolded?
               What events occurred during this time frame that are most relevant to the plot(s)? 

 

Time Frame

use month/years e.g. June 1789 – April 1942

 

Provide a text list of events following the sample layout.  Use short description (i.e. one or 2 lines per event)

Provide what you think is the best subset of events (20 events MAX)

 

	Event	Date Can be a range	Most relevance source files (5 Max)
1	Von Ryker Institute shut down in Switzerland	Fall of 2001	Scientific Review 4
2	BSE found in Mabton	December 2001	1101243512780
3	Boynton Laboratories founded	May 2002	1101163356450
4	Boynton Laboratories receives contract for BSE testing	September 2003	1101163018612
5	John Torch enters mayoral race	July 2003	1101631275108
6	Pictures taken of John Torch with a woman that is not his wife	April 2004	1101163356001, picture084, picture085
7	John Torch essentially drops out of mayoral race	Summer 2004	1101162945890
8	Rex Luthor pulls ahead in mayoral race	Summer and Fall 2004	1101162945890
9
10
11
12
13
14
15
16
17
18
19
20 max

---

3. WHERE: What locations are relevant to the plot(s)?

Follow this example layout.  Use only one-line per item.

	LOCATION	Most relevance source files (5 Max)
1	Switzerland	1101163356001, Scientific Review 3, Scientific Review 4
2	Boynton Lab in Alderwood	1101163356450
3	Mabton	1101243512780
4	South America	1101163977242, 1101163018599
5 max

 

---

4. DEBRIEFING

The founders of Boynton Labs arranged for a cow in the Alderwood area to be infected with mad cow disease. The motivation behind this was to build up support for creating their new lab within the community and within the local government. The discovery of mad cow disease in the US also instigated changes in the testing for mad cow disease which directly benefited the lab. Dr. Alejandro VonRyker and Dr. Delwin Sanderson wanted to create a situation that would allow them to continue their controversial research on mad cow disease under the cover of a legitimate lab. Previously in Switzerland, they had been involved in running a lab conducting research that included purposing infection cows with prion diseases. It was suspected that Dr. VonRyker also deliberately faked two research papers on synthetic transmissible prion diseases. The Swiss authorities shut down the institute in Switzerland in late 2001 shortly before mad cow disease was then discovered in the US near Alderwood. Mayor Rex Luthor is a staunch supporter of the lab because of the business it is bringing into Alderwood and the benefits he receives. Laurel Sulfate is also a supporter of the lab and helped to discredit John Torch by having coffee with him at a local Starbucks. Pictures were taken of the two together and it was implied that John Torch may be having an inappropriate relationship with this woman considering that he is married with children. John Torch was a candidate in the mayoral race against Rex Luthor so Rex most likely used his influence with the lab to set up the scenario. It is unlikely that Torch was actually having an affair because the photos taken did not indicate anything inappropriate and because Torch seems to be a man of firm beliefs. Testing of some of the research in the lab is being carried out in South American countries.

 

 

5. VISUALS and Description of ANALYTIC PROCESS

The first thing we did was to try and get a feel for the depth and scope of the data set provided. We reviewed all data provided in detail with the exception of the news stories. For the new stories, we wrote a script to rename the files based on the published date as shown in Figure 1 below. This allowed us to skim some articles in chronological order. Again, the intent was to familiarize ourselves with the data.

 

 

Figure 1: Articles after being renamed based on published date.

 

This process led to the hypothesis that the outbreak of mad cow disease was not an accident, however we did not have any further details. We then proceeded to design an interface we believed would highlight the links among what we defined as the five primary data types: individuals, organizations, events, activities, and locations. We focused on a creating a visualization that would assist a user in discovering relationships among the data elements in the data set. This approach led us to decide to use “classic” visualization techniques combined in a novel way. We chose to use a tree map, timeline, map view, network visualization and lists of all the entities. Once one or more entities are selected and then the Filter option is selected, all views update to reflect the user’s choice. As an example, figure 2 shows the visualization with data selected and filtered.

 

 

Figure 2: Example data filtered in the visualization. The person “Rex Luthor” and the activity “2004 election” are selected for this view.

 

To come to our conclusions about the data set, we used the tool to select and then filter on information about specific items we thought were of interest. In particular, we started with mad cow disease and used the network visualization to see connections to other entities, the timeline to track story headlines, and the actual list of articles to open and read for further information. We opened the articles by double clicking the articles of interest in the list (see Figure 4). Figure 3 shows the application with mad cow disease as the filtered entity. We believe one of the strengths of our tool is that it helps reduce the number of articles of interest from the almost 1200 provided to just those that appear based on the user selected filters.

 

 

Figure 3: The visualization with mad cow disease as the selected filter. The links to other entities can be seen in each of the other components of the visualization above.

 

 

Figure 4: Example news article selected by user.

 

We reached our final hypothesis through directed exploration of the data using the tool. Specifically, mad cow disease led to further exploration of Rex Luthor, Dr. Delwin Sanderson, Dr. Alejandro VonRyker, Dr. Philip Boynton, and Boynton Laboratories. It also led to the exploration of the source of the cow that tested positive for mad cow disease. Looking into the Alderwood City Council came up from the above explorations which led to the analysis of John Torch and the mayoral election. We believe he was being set up because in the supposedly steamy pictures, he is shown with a Boynton Labs employee having coffee but not being inappropriate. His record on voting against liquor licenses and his condemnation of Dr. Martin Luther King Jr. as being an adulterer led us to believe that he is a man that stands up for his values (popular or not) and is unlikely to have had an affair himself. The timeline view assisted in this conclusion. It clearly displays the progression of events.

 

Figure 5: Timeline of John Torch with one of the later article details displayed from a mouse over.

 

 

The strength of the relationships between entities can be shown with ease using the node-link graph. Figure 6 below shows the connections between Rex Luthor, Boynton Laboratories, and mad cow disease. Multiple connections between entities, especially if they were unexpected, lead to further analysis of the nature of the connections.

 

 

Figure 6: Node-link graph with Rex Luthor, Boynton Laboratories, and mad cow disease.

 

 

Admittedly, this method does take time and sometimes leads down the wrong path. However, it provides solid direction and allows the user to make informed decisions about what to explore next that seems most promising.

 

TOC:  Who – What – Where – Debriefing - Process