Basic exploration features
The main view comprises thumbnails of Solution Space Visualization
Explorer papers that are positioned according to their mutual
dissimilarity, i.e., the thumbnails located near each other
correspond to the papers with low dissimilarity/distance (see
below for details).
By clicking on a thumbnail in the main view, you can open a dialog
with detailed information on the paper title, corresponding types
and properties of data, supported visualization and analysis
tasks, corresponding publication, and a BiBTex file with
bibliographical information.
To explore cluttered regions of the main view, it is possible to
pan it by dragging with the pressed left mouse button, and to zoom
it with the mouse wheel.
Hovering over a technique thumbnail will blur the thumbnails for
dissimilar papers, and display the links to similar papers as well
as the actual similarity value normalized in range [0; 1] (0
meaning minimum similarity / maximum distance, and 1 meaning
maximum similarity / minimum distance). See below for more
details.
Right-clicking a paper thumbnail will select it and provide an
effect similar to highlighting on hover, except for the fact that
such selection will be persistent until the same or other
thumbnail is right-clicked.
Basic search and filtering
By using the panel in the left side of the page, it is possible to
search for specific papers (the query string will be searched for
in all fields such as the publication information and categories)
and to filter the displayed papers based on the publication year
and the assigned categories.
Advanced features
The "Advanced" tab in the left side of the page provides controls
for re-adjusting the parameters related to similarity
computations.
By changing the values for specific similarity factors, it is
possible to change their weight when computing the similarity for
each pair of papers. For instance, setting a value to 0 will
essentially exclude the factor from computations. After each
change, the layout in the main view is recomputed using the
updated combination of factors.
The factors include the following:
- Difference in the publication year values
-
Difference in the assigned sets of categories (calculated
using Jaccard index)
-
Difference in the sets of authors parsed from publication
information (calculated using Jaccard index)
By changing the value for the similarity cutoff/threshold, it is
possible to control which papers will be considered
similar/dissimilar during interactive exploration (highlighting
and selection described above). For instance, setting this
threshold value to 0 will essentially consider all papers similar
to each other (and the interactive highlighting will make no
sense). By setting the threshold larger than the default value, it
is possible to focus only on more similar papers.