KnowledgeSpaceVisualizationTool

From AnalysisWiki

A demonstration of the Knowledge Space Visualization Tool is available as a Java Web Start (JNLP) file:

NEW: Semantic field visualization (November 6, 2007)

Summer Institute 2007 Abstracts Demo (you need Java 1.5 or better to use this).

NOTE: There is a bug that prevents the visualizer from appearing until you resize the window. It's a java layout problem that I'm having trouble tracking down. Sorry!

A screen shot is shown below. Some notes:

• The boxes on the left are zoomable: click in the box and use your scroll wheel (Command-scroll on a Mac, or double-finger scroll on a new MacBook) to zoom in and out. You can pan the area by clicking-and-dragging the area (bottom) or holding down the ALT or Option key and clicking and dragging (top).

• You can highlight a group of notes in the top box by clicking-and-dragging.

• Once something is selected, you can select "Analyze" to see some summary statistics about your selection(s) in the "Details" area.

• Once something is selected, you can select "SF+" to generate a semantic field in the lower display. You can then select something else and click "SF+" again to overlay another semantic field. This can be repeated indefinitely. Use "SF-" to clear all semantic fields.

• The semantic field extraction algorithm in this version uses the Yahoo! Term Extraction service. A purely LSA-based term extractor will be made available soon.

• Notes (documents) are shown as blue squares (annotations are yellow), people (authors) are shown as purple triangles, views (folders) are shown as grey squares, and scaffold supports are pink diamonds. Link colours generally correspond to components except for linkages between notes, which are coloured according to relationship: blue = build-on, magenta = reference, yellow = annotation. Semantic relationships between notes are shown as red lines.

• Note icons (the squares) can be scaled by word count, latent semantic vector length (a measure of shared knowledge content), a combination of those two, the number of times they have been revised and the number of times they have been read.

• Hovering over an icon will show a "tool tip", which in the case of a note is a brief overview of the note.

• Double-clicking on a note icon will show the note contents in the rightmost panel

• An overview is always available in the upper middle box.

• You can use the date slider (drag the boxes at the end of the slider toward the middle to constrain the date) to constrain the notes shown. Creation dates (rather than modification dates) are used for this query.

• You can enter search criteria in the boxes labeled 'title', 'author' and 'text'. Words within a box are joined via the OR operator. Box contents are joined via an AND operator. You must press the 'Return' key when you are finished entering text in a box. Clicking "zoom to fit" will zoom the main panel to include all highlighted terms.

• 'Links' refers to the choice of what type of link to turn on, both for display and for determining what affects the force-directed layout. The slider beside "semantic" refers to the minimum semantic similarity needed for nodes to be joined when semantic links are enabled. Lower values will result in more linkages being shown. Another way to think about this is that it refers to the granularity of similarity. _NOTE:_ for the demo data set, use a cosine of 0.1 for interesting results.

• 'Start' and 'Stop' determine whether the force-directed layout is allowed to find an optimal solution.

Questions or comments should be directed to Chris Teplovs.

Reference: Teplovs, C. and Scardmalia, M. (2007). Visualizations for Knowledge Building Assessment. Paper presented at the AgileViz workshop, CSCL 2007.