Discourse Trees

[kayaulai]

What to do

1. Implement layers for trees.
2. Add a new feature 'Create a Discourse Tree from stack', with units instead of tokens as leaves.
3. In the tree pane, show a drop-down menu to let the user choose between:

• syntactic tree
• discourse tree

4. When the discourse tree option is selected:

• In the right pane (tree pane), let the user click on a unit to add it to the current discourse tree.
• (Use the same UI that already exists for making sentence trees by clicking on tokens [= dependency syntax].)

5. For the discourse tree pane (right pane), provide two rows at the bottom of the pane:

• upper row displays a series of boxes/buttons representing the units in the stack (showing just the first 10-15 characters of the unit) [This is similar to the existing tree pane row, which displays a series of tokens.]
• lower row displays the entire (concatenated) text of one unit (= the user's currently selected unit)

6. Alternatively, we could add multiple (scrollable) rows at the bottom of the discourse tree pane, one row per unit in the stack.
7. The trees will have "nodes" with many words; to display them:

• truncate (with a ... in the bottom line), or
• Implement word wrap (too complicated?)

9. If more space is needed, the tree pane could be expanded to fill the full width of the screen (and even the full height).

Background (For Jack:) I believe the current tree format can be easily modified to better support RST and CA:

For CA, since we already allow roots, e.g. first-pair-parts (FPPs) and second-pair-parts (SPPs) can be the root of a tree; then we can have e.g. a pre-sequence and a post-first repair sequence attached to the FPP, a pre-pre to the pre-sequence, etc.

See also #503 #507