# Leopar User's documentation

• Graphical User Interface

# The Graphical user interface

This page describes the documentation of the GUI of Leopar. Leopar is started with the command

leopar

## The main window

After few seconds for lexicon and grammar loading, the main window is available:

From this window, you can:

### Parse a sentence

The button Parse runs the parsing on the sentence given in the text entry above (Jean voit la fille avec un télescope. in the image above).

It is possible to choose sentences from an examples file with the combo list. The current examples file name is shown on the button below the text entry (examples.txt in the image) and it can be change with the same button. The + and - buttons allow to add or remove examples from the current examples file.

If Leopar succeeds, the result of the parse is displayed in the result window (see below).

An interactive mode for the current sentence is available (<font color="red"><b>Interactive parse</b></font> button) for parsing. There are no specific documentation, your are invited to try it and to blame us if it is not easy to use.

The two last buttons (Tokenize and Filter) are mainly useful when Leopar fails to parse a sentence: it is possible to have different views and to understand where the parsing process fails.

### Consult the lexicon

From the main interface; it is possible to consult the current lexicon: for a given word (like voit in the image above), the button with a magnifying glass above open a new window with the set of hypertags that the lexicon associates to the word. Hypertags are organized in trees where entries with the same lemma are grouped.

### Consult the grammar

There are two ways to consult the grammar:

1. You can consult the whole grammar (the <font color="red"><b>Browse</b></font> button)
2. You can ask for the set of EPTD that are linked to some word: use the last text entry (like <font color="orange"><i>fille</i></font> in the image above) and the button with a magnifying glass above. A new window, is displayed with a tree structure like in the lexicon case but with one more level containing the anchored EPTDs.

## The result window

When a parsing succeeds, a new window is opened:

The window is separated: on the right, there is the main display window and on the left a tree view can be used to change views on the right.

There are 3 different modes for viewing parsing results:

1. with the <font color="red"><b>Complete dependencies</b></font> button, the set of solutions are clustered: different trees which produce the same dependency structure are in the same subtree and nodes of depth 1 in the tree view are the dependency structures. The image above correponds to this view.
2. with the <font color="red"><b>Epsilon dependencies</b></font>, the set of solution are also clustered, like in the previous case but with a different dependency structure representation (see Dependency extraction in a DIG grammar for detail about complete and epsilon dependencies
3. with the <font color="red"><b>Parse trees</b></font> radio button, no clustering is done and the set of parse trees that are parse structures for the input sentence are displayed at depth 1 in the tree view.

Note that if you want to use Leopar with a grammar which is not a DIG grammar (in the sense of DIG grammar), then only the third mode will be available.

## Deeper in the result window

Either with the <font color="red"><b>Parse trees</b></font> radio button or by tree exploration in the dependency view mode, you can access to the parse tree view.

This view contains several tools to help to understand how the sentence were parsed. The blue line in the image abobe contains thee small icons:

1. Up arrow produces a in depth view in the display tree and show dependency and parse tree view at the same time
2. Paper and pen links to an interactive windows where the whole deep parsing is available (first image below)
3. Paper and try square links to a view of the list of initial EPTDs used for the current parsing (second image below).