Thumbnails xLDD application

© 2011 Europaeische Akademie Bozen/Accademia Europea Bolzano All rights reserved.

Chris Culy, Verena Lyding, and Henrik Dittmann

Show explanation

Hide explanation

This is an application for the user-driven visualization of dependency structures as Extended Linguistic Dependency Diagrams (xLDDs). Two small corpora can be searched: a selection of the Italian PAISÀ corpus and a set of news releases from EURAC in German. The Italian corpus is parsed by the DeSR dependency parser from Pisa and the German corpus is parsed with the dependency parser of the mate-tools. To search the corpora, choose one from the dropdown menu, then type in a CQP search phrase. Word searches just require that the individual words be in quotes. e.g. "von" "dem". Word searches are case sensitive. You can search for words that have a dependent relation, e.g. direct object: obj (for Italian) and OA (for German), like so:

		[deprel = "obj"] (for Italian)
		[deprel = "OA" ] (for German)

You can search for words that are the head of a relation (e.g. verbs that have direct object), like so:

		[deps contains "obj:.*"] (for Italian)
		[deps contains "OA:.*"] (for German)

The search results are the sentences containing the search terms. The search results are displayed in a table below. Each line will show a small version of the dependency structure for the corresponding sentence. Holding the shift key down while double clicking on a small version will show a larger version, and holding the shift key down while double clicking on the large version will return to the table view.

You can specify and alter the visual encodings of text and arcs according to the dependency relations and additional information associated with tokens of the dependency structure. The encodings are applied to all diagrams by clicking on "apply visual encodings" button. The encoding alternatives are limited to a predefined subset of visual encodings offered by the xLDD tool. This results in an easy-to-use but yet fairly flexible tool. For an example of more detailed settings, see the PAISÀ example.

The diagrams allow you to encode information in various ways to make it easier to see patterns quickly.

For the encoding of dependency relations you can specify one or more dependencies and assign to them an encoding. As well you can choose an encoding for the unselected relations. The options for encodings of dependency relations are:

Token level information can be encoded for words, lemmas and parts of speech (pos), by naming them in a comma separated list. Encoding options for token information include:

You can explore the diagram in a variety of ways:

Search term:

Specify visual encodings:
Show arcs for dependency relation(s) (items separated by commas) as and all other arcs as

Show tokens of the (items separated by commas) as



Adjust diagram size: