Report for Design Challenge 1 - CS 838 - 2

According to the interview and details described by the domain scientists, the major difficulty in the domain research is to determine the relationships between Epistemic Frames. Thus, in our opinion, the most important purpose of our design is to clearly show the comparison of frame matrices while providing detailed comparisons for each element in the matrices if needed. Therefore, we decided to create a hierarchical multi-interface application written in C++ including three layers of view for the frame data. The first layer of view was an attempt to show distances and clustering of frame data (matrices) with respect to a reference (a matrix chosen by the users). By clustering the frame data, this application uses distance to compare the significance of differences between peripheral frames to the center (reference matrix). The longer the distance, the larger the difference. According to the domain scientists, the absolute value of length may not be very meaningful, but the clustering shows which groups of data frames are close, i.e. probably implying closer relationships or correlations. Our initial difference function simply sums the absolute values of the differences between each element of a given frame. Our design provides an interface for users to define their own distance functions so that the effects of different distance measurements can be explored. If the user is interested in the relationship between two elements, he is allowed to select the specific matrices and look into their differences element by element with the second and third layers of the view. The second layer of view is an attempt to show distribution of differences in two frames of data. We decided that color saturation might be an efficient way to represent the absolute difference of each element. For each pair of frames (matrices), we create an identical matrix brick diagram and map the color saturation of each brick in the diagram to the range of difference between the corresponding elements of the target matrix element pair. This feature gives the viewer a quick visualization of the distribution of the differences between each element. The third layer of the view was a direct comparison of each element of each matrix selected from the first layer. Similar to the second layer, a matrix brick diagram is created, but now each brick is used as an area for a bar chart generated from the value of the corresponding elements of the selected frames. This view gives the audience a detailed understanding about how and which parts of the several data frames are different. It allows multiple comparisons for more than two data frames, which is advantageous.