lo-fi prototype
Milestone date: 18-Feb-2009
Based on our understanding of the target audience and the data, there were several questions that we hoped a user could answer with our visualization:
- How do the texts in the CCEL relate to the bible?
- How do the texts relate to each other both in terms of how biblically-based they are and what portions of the bible they focus on?
- How do authors' use of scripture change over time?
- Where can I find texts about a particular scripture book or verse? What might the contents of these books be like?
conventions & data profile
Building off of our audience description, we recognized several conventions that our audience might recognize that we can take advantage of. Our users are very familiar with the Bible, including the relative linear placement of biblical books (e.g. a typical user could open a Bible directly to a given book with very few page flips). Furthermore, they are likely to categorize biblical books according to a relatively standard schema. A Google image search for "bible books" returns many related images that reveal this organizational schema:
 |
 |
 |
 |
 |
Recognizing this convention is import because one of the primary concerns of our users is how a book relates to scripture. This also suggests that one of the most useful aspects of our data will be the scripture references that exist in many of the CCEL books. Looking at the data, 672 of the 868 books available have 1 or more marked scripture reference. Of these, the average scripture reference count per book is 1,083 (692 excluding outliers) but the median is only 180, indicating a heavily right skewed distribution. We also previously highlighted the importance of a book's relationship to history. Among the information available in the CCEL texts is reasonably complete information about the author life span. While original dates of publication are often not available, this can serve as a proxy for this information, placing an author's books at a particular range in history.
prototype
We choose a visualization that shows the relationship between the books in the CCEL and the Bible in a way that also reveals relationships among the CCEL texts and demonstrates how scripture is referenced over history. This is done by placing representations of CCEL texts spatially based on the number of references to the Bible as well as by the distribution of biblical books cited.
The x-axis is a representation of the Bible based on our observed convention, where each genre is color-coded. If a CCEL text cites one biblical book exclusively, one would expect it to appear directly above that book. If it cites two biblical books equally, it would appear in the horizontal middle between those two books.
The y-axis is a logarithmic axis of the scripture reference count, ordered from top to bottom. A logarithmic axis is used due the right skew of the data. The counter-intuitive ordering was chosen for two reasons: First, books that are high in citations will likely have many edges (discussed below) and the reversed ordering reduces overlap. Secondly, it metaphorically matches the concept of a text being "biblically-based" or closer to our representation of the Bible (the x-axis).
In addition, edges are drawn between the biblical books referenced and the CCEL text that contains the reference. The edge thickness hints at the relative distribution of references. While it is not strictly representative, in order to reduce clutter, this information is dual-encoded with the x-axis placement. Furthermore, these edges reflect the genre of the biblical book cited.
Hopefully, arranging the texts in the CCEL in this way will reveal emergent patterns. For instance, the above sketch includes the profile of several biblical commentaries. A book like William Newell's Romans Verse-by-Verse with 2,670 scripture references would hopefully appear immediately above the biblical book of Romans with a thick line pointing straight down from it. Books that are about New Testament topics in a broad-sense might appear in the upper right portion of the graph with a pyramid-shaped edge profile.
looking forward: interactions
The above visualization suggests several possible interactions. One that we imagine, but that is not conveyed at this level of fidelity, is the idea that the edges would have a slight-degree of transparency by default. Hovering over either a CCEL text or a biblical book would remove that transparency from the associated edges, causing them to pop-out against the background.
As the above sketch suggests, a graph containing all 672 texts might be impossible to decipher. Furthermore, we would like to answer the question of how patterns of scripture reference change over time. Therefore, the main interaction we designed is a time slider which can filter the books displayed to a certain historical window.
This slider contains a Gantt-like chart of the authors life span (and reveals the relative historical distribution of CCEL texts). The control window would be both movable and of flexible size. Dragging the window would cause books associated with the authors represented to fade in or out from the visualization, providing a visualization of the changes in scripture reference patterns over time.
An outstanding challenge is to design a way to zoom in on a particular biblical book in order to observe the references to a verse of interest. Such a view should strive to make the most of screen space while preserving some of the context of the visualization shown above. Finally, users should be able to use this visualization to navigate to the actual text of the CCEL documents; however, we do not see this visualization's primary purpose as a navigational tool.
user feedback
We showed the above visualization to a number of people (6-7) informally in order to collect feedback. While not all of these people fit the profile of a CCEL user, they all exhibited generally positive feedback while highlighting some areas for potential improvement.
The most enlightening test procedure was to ask participants for their initial reactions without any explanation of the visualization. It was clear that this proved very difficult for all but one user, and revealed that it is not immediately clear what the nodes and edges in this visualization represent. Users familiar with the Bible immediately recognized the x-axis representation, but often wondered if the term "references" referred to cross-references within the Bible (e.g. Chris Harrison's Visualizing the Bible). Furthermore, one user said that she found the numbers on the y-axis to be confusing. This suggests that some careful redesign of the y-axis, in order to better establish context, will be necessary. Furthermore, a more iconic representation of the CCEL texts might help with the identification of all the elements present.
The user who understood the visualization without any explanation demonstrated this by asking questions like "So, are the books sort of pulled into the horizontal position based on the distribution of references in that book?" Furthermore, he was able to correctly identify the profile of certain types of books, such as scholarly commentaries and broadly-focused books on Christian living. Other users also demonstrated this ability after some explanation of the elements involved. One of the users originally interviewed for our audience description had this to say: "Overall, I think it conveys a lot of information about the books in the CCEL and their use of scripture in a succinct visualization, which is no doubt what you're going for. It does take a bit of mental energy to interpret it, but I think that's true of most visualizations." Hopefully, with some careful refinement, we can reduce this cognitive load while still conveying rich information about the CCEL.