From CWRC


Revision statistics per user
User Total #
of edits
Sentences Links
Owned Proofread Deleted internal external
current total current total sentences releases current total current total
DanielSondheim 7 15 0 0 0 0 0 0 0 0 0
JanaSmithElford 81 159 0 0 0 0 0 0 0 3 3
SusanBrown 4 5 0 0 0 0 0 0 0 0 0
SusanHesemeier 8 5 4 0 0 0 0 0 0 0 0
-- 0 4 (2.13%) 0 0 0 0 0 0 0 0 0
    
Page statistics (latest revision)
StatisticValue
Edits (last 30 days) 0
Edits (total) 100
Incoming Internal Links 2
Page Views 35487



Ovis

Contents

Introduction

OrlandoVision (or OVis) is an experimental, interactive link-node graph designed for exploring the Orlando textbase. It enables the user to visualize the writers with entries in the Orlando textbase and the more than 30,000 people associated with them.

How Ovis Works

OVis represents the connections between people as nodes and the relationship between them as edges, and enables the user to read about the context of the link through the text reading pane. The links or edges are colour-coded, which corresponds to the colouring of the particular semantic tag that indicates the context in which the association is mentioned, such as <politics>, <textualfeatures>, or literary <responses>. The tool also has a multi-functional search capacity which enables users to navigate large swaths of data at once. Users can learn about the connections between individuals in the Fabian Society, for instance, by searching for "Fabian" and tracing the links between the nodes (people) who appear on the graph.

Ingredients

No external ingredients are needed.

Steps

    ===Open and Load===
  1. Double-click the Ovis icon on the desktop.
  2. Under "File," select “Open Data.”
  3. Navigate through the folders by clicking on "Desktop," "CWRC," "ovis," and "DATA" until you find a file called “orlando_entries_all.” Double-click it.
  4. Wait briefly for the dataset to load.
  5. Navigate

  6. In order to navigate within the graph itself, you will need a conventional mouse.
  7. To zoom into and out of the graph, use the scroll wheel, or right click and drag.
  8. To move the entire graph, click the scroll wheel, hold, and drag into desired position.
  9. To highlight (or select) a node or edge, left click on a node or edge.
  10. To highlight multiple nodes or edges, press down the shift key and left click additional nodes and edges, or left click and drag a rectangular lasso.
  11. Once a node is highlighted, the person's name will appear in the text pane at the bottom of the screen.
  12. Once an edge is highlighted, the text extracted from the Orlando database will describe the context of the link in the text pane.
  13. If you select multiple edges, the text extracted from the Orlando database will describe all the links you have selected. This text will appear in the scrollable text pane. The most recent links you have selected will appear at the top of the text pane.
  14. To deselect nodes and edges, left click any empty space in the graph.
  15. Search

  16. In OVis, you can search using any of the following search options: "Text", "Author" (writer), or "Date Range".
  17. To use the "Text" search feature, click on "Set" next to "Text" on the right-hand side of the screen.
  18. Click “Add” in the window that pops up.
  19. Double-Click in the empty box under “Search Term.”
  20. Type a search term in the text box.
  21. To enter multiple search terms, click "Add" and enter another search term.
  22. Click the box under "And/Or" to select "And" or "Or".
  23. Click the box under "Stem" to select whether to stem your search term and include all variations of that term.
  24. Select whether you want your search to be "broad" (showing all writers with entries in the Orlando database that mention the search term AND all other people to whom they are connected) or "narrow" (showing only the entries in the Orlando database that mention the search term).
  25. Click "Remove" to remove the bottom search term.
  26. Click "Cancel" to cancel search.
  27. Click "OK" to set your search.
  28. To use the "Author" (writer) search feature, click on "Set" next to "Author" on the right-hand side of the screen.
  29. Click “Add” in the window that pops up.
  30. Double-Click in the empty box under “Search Term.”
  31. Type an author name in the text box.
  32. To use the "Date Range" search feature, click on "Set" under "Date Range" on the right-hand side of the screen.
  33. A window will pop up.
  34. Choose one of the options under "Historical Dates" or "Monarch Dates". DO NOT select both of these options.
  35. Alternatively, enter a Start year into the text box below "Start" and "Year (1+)", select a Month and Day, and enter a End year into the text box below "End" and "Year (1+)", and select a Month and Day.
  36. Click “OK.”
  37. Limit Results

  38. Limit your results in OVis by selecting or deselecting the following options from the right-hand side of the screen below "Display Properties": "Visible Nodes", "Writer Sex" and "Writer Type".
  39. Results may also be limited by selecting or deselecting Tags, on the lower right-hand side of the screen below "Active Tags".
  40. To select all tags, click the "Check All" button.
  41. To deselect all tags, click the "Uncheck All" button.
  42. To select or deselect a particular tag, click the box to check or uncheck that tag.
  43. Additional limiters are available through the "Search" function.
  44. Modify Layout

  45. To change the size of nodes and links (edges), use the sliders on the top right of the screen.
  46. To change the node colour, click "writers with entries" or "other people". Select colour from the colour wheel. Click "OK".
  47. To change the node style, click on the "View" menu and navigate to "Set Node Style", which has more customization options.
  48. To change the background colour, click on the "View" menu and navigate to "Set Background", which has more options.
  49. To change the text that appears in the upper left hand corner of the screen, click on the "View" menu and navigate to "Set Annotation". A text box will appear. Write in your desired annotation.
  50. To eliminate the text in the upper left hand corner of the screen, select or deselect "Show Annotation" in the "View" menu.
  51. To see different ways to display the graph, click on the “View” menu, navigate to “Set Layout Strategy,” and select one of the other options. (**For a more detailed exploration of the Layout Strategies, see "Features" below.)
  52. Save and Reload

  53. To save a copy of your OVis Session, select "Save Session As" from the "Session" menu.
  54. Write in the name of your session in the textbox that appears.
  55. Select a folder in which to save your session.
  56. Select the "Save" button.
  57. To save subsequent changes to your session, select "Save Session" in the "Session" menu. This will save any subsequent changes you make to the session under your original session name.
  58. To revert to your saved session once you have made changes to your graph that you wish to remain unsaved, select "Reload Session" in the "Session" menu.
  59. Once you are finished with your session for a time, quit OVis.
  60. To reload your previous session, double-click the OVis icon.
  61. Under "Session," select “Open Session.”
  62. Navigate through the folders until you find the file you saved. Double-click it.
  63. Take a Screenshot

  64. To take a screenshot of your graph, select File from the menu and select Screenshot.
  65. Save your screenshot where you want it.
  66. To make the resolution of your screenshot higher, select "view" and then "Set Snapshot Magnification". Select desired magnification (from 1-10).
  67. This screenshot feature will only take a picture of the graph itself, without the OVis frame.
  68. To take a screen shot of the entire screen, press CTRL+SHIFT+4 on a Mac Keyboard, or the PrtSc button on a PC.

  69. Try the example and suggested activities below!

==Features==


Layout Strategies

The layout strategies available in Ovis can be divided into two categories: Force-Directed and Non Force-Directed.

Force-Directed

In the case of the Orlando dataset, Force-Directed layouts will generally prove to be more useful, at the cost of requiring more time to produce results.

Force-Directed layouts use a physics based approach to arrange the nodes of a graph. The general goal of a force-directed layout is to attempt to ensure that the edges, i.e. the connections between nodes, are similar in length, and that they do not cross. However, given the structure of the graph this may not always be possible.

The best way to understand a force-directed layout is to think of the edges of a graph as though they were springs, and the nodes as though they were electrically charged particles. The nodes repel each other, while the springs attract. The interplay of forces continues until eventually the system reaches a point of relative equilibrium. At this point the graph is drawn using these positions.

For the most part, the force-directed layouts described below tend to represent the nodes with the most edges between them as being closer together in the graph. This is because the more edges between nodes, the stronger the attractive force between the nodes. This is particularly true of the "Simple 2D" and "Force-Directed" Graphs, which may prove the most useful for the Orlando dataset, at the cost of requiring more time to produce results.

For a more detailed explanation, please see: http://en.wikipedia.org/wiki/Force-based_algorithms_%28graph_drawing%29


Simple 2D

This layout uses the Fruchterman-Reingold algorithm (https://nwb.slis.indiana.edu/community/?n=VisualizeData.Fruchterman-Rheingold). Like the ForceDirected algorithm, it is fairly slow. A unique feature of this algorithm is that it uses an overall “temperature” in order to guarantee that the system will reach an equilibrium. This algorithm tends to produce graphs which are more dispersed than the ForceDirected algorithm.


Force-Directed

This layout uses the Kamada-Kawai algorithm. It is a slow, force-directed layout algorithm and tends to produce graphs which have more clusters than the Simple2D algorithm.


Clustering2D, Community2D, and Fast2D

These three layout algorithms are all fairly similar. While Clustering2D tends to produce a graph with more crossed edges, Community2D and Fast2D produce almost identical results. These algorithms create graphs much more quickly than the previous two layout algorithms, because they use a density grid which provides a random initial rough layout of the graph.

Non Force-Directed

Non Force-Directed layouts tend to be more random than Force-Directed layouts. This is particularly true of the "Random," and Circular layouts, which both place nodes within the boundaries of a graph in a random pattern.


Random

Random is an extremely fast layout algorithm. As might be expected, this layout randomly places nodes within the boundaries of the graph. It is therefore not very helpful for displaying meaningful connections; however, due its speed it could be useful on slower computers and/or when working with a large number of nodes.


Circular

This is a fast layout algorithm which arranges nodes around the circumference of a circle. Since there is no logic applied to the positioning of the nodes, it is not very useful for displaying meaningful connections.


Span Tree

This is a fast layout algorithm which arranges nodes into a series of circular spanning trees. A spanning tree can be defined as the minimum number of edges that connect all the nodes of a graph. In the case of the Orlando dataset, we are usually working with a spanning forest, which consists of multiple, separate spanning trees. This layout is best suited to sparse graphs, wherein the number of edges is about the same as the number of nodes. For more information on spanning trees please see: http://en.wikipedia.org/wiki/Spanning_tree

Example 1 (please refer to the above instructions while trying these examples)

  1. Open and load the Orlando dataset, following the instructions above.
  2. Using the "Text" search feature, search for "theosophy" OR "theosophist" (both unstemmed).
  3. Limit your search by selecting "Narrow" (showing matching links only) in the "Text" search feature window.
  4. Click "OK" to initiate this search.
  5. Wait for the graph to load.
  6. Set the "Fast 2D" layout strategy from the "View" menu.
  7. Wait for the graph to load.
  8. Determine what node is at the centre of the graph by using the navigation features of your conventional mouse.
  9. Estimate which link or tag colour is the most prevalent in the graph, using the navigation features of your conventional mouse.
  10. On the lower right-hand side of the screen below "Active Tags", click on "Uncheck All". Then select one tag to view only one type of link (ie: the "politics" tag).
  11. Determine which nodes have the most links between them, using the navigation features of your conventional mouse.
  12. Highlight one edge (or link) on the graph, by clicking on that edge using the left mouse button.
  13. Read about the context of that link between two nodes in the text pane at the bottom of the page.

===Example 2===

  1. Open and load the Orlando dataset, following the instructions above.
  2. Using the "Text" search feature, search for "Fabian Society" (unstemmed) OR "Fabian" (stemmed).
  3. Limit your search by selecting "Narrow" (showing matching links only) in the "Text" search feature window.
  4. Click "OK" to initiate this search.
  5. Wait for the graph to load.
  6. Set the "Force-Directed" layout strategy from the "View" menu.
  7. Wait for the graph to load.
  8. Determine what nodes (people) are at the centre of the graph by using the navigation features of your conventional mouse.
  9. Using the shift key, select these nodes (people).
  10. Determine which nodes have the most links between them, using the navigation features of your conventional mouse.
  11. Estimate which link or tag colour is the most prevalent in the graph, using the navigation features of your conventional mouse.
  12. Using the shift feature, highlight the edges (or links) that connect the nodes (people) in the graph.
  13. Using the shift feature, draw a path between these people.
  14. Read about the context of the links in the text pane at the bottom of the page.

==Suggested Activities==

  1. Try to find what connections exist between people involved with the "Men and Women's Club". Restrict the graph to "narrow" to get a sense of these connections. Observe what tag colours seem the most prevalent in the graph. Assess what figures appear to be central to the club by selecting the "Force-Directed" layout strategy, then viewing the nodes at the centre of the graph. Click on the links between individuals. Read about the context of these links.
  2. Look at all people connected to the "poetess" movement who were alive during the early Victorian period. Restrict the graph to writers with entries to get a sense of relationships between them. Deselect all tags and then select only the reception-oriented tags to get a sense of the relationships involving reception amongst the group. See if you an arrive at a sense of who were the key figures related to reception in this literary network.

==Discussion==

It should be remembered that this is an experimental interface that has not yet been officially released. This means that some bugs can be expected.

Description

A document with a more general description of OVis is available here: http://entry.tapor.ca/?id=5

Who has worked on creating it

Susan Brown guided the prototype design from a literary historical perspective, Michael Bauer directed the technical work, Jennifer Berberich and Jonathan Cable did the coding of the first prototype, and Patrick D. Emond coded the second prototype. Jana Smith Elford did significant testing and drafted the documentation.

Work on OrlandoVision has been generously supported by the Sharcnet High Performance Computing Consortium, which provided programming resources, by the Social Sciences and Humanities Research Council of Canada, and by the University of Alberta’s Faculty of Arts.

Where to read about it

Smith Elford, Jana, Susan Brown, Michael Bauer, Jennifer Berberich and Jonathan Cable. “Elevating Influence”: Victorian Literary History by Graphs." The Victorian's Institute Journal Digital Annex. 1 (1). 2011.

Click here to return to Workshop Homepage