STARK

Welcome to the online demo interface for STARK - a highly-customizible tool designed to extract various types of syntactic trees from dependency-parsed corpora (treebanks). Unlike the original command-line version, this user-friendly interface offers a streamlined set of settings, which are described in more detail here. Simply upload your treebank and click ‘SUBMIT’ to view the initial results!

Input data

Upload a treebank in CONLL-U format

Browse

{% if 'file' in validation %} {% endif %}

Or insert a URL link to a treebank in CONLL-U format {% if 'input_url' in validation %} {% endif %}

Tree specification

Tree size: number of tokens in the tree ()

Node type: token characteristics to consider ()

Part-of-speech {% if 'node_type' in validation %} {% endif %}

Lemma

Form

{% if 'node_type' in validation %}

{% endif %}

Advanced settings

Labeled trees: include names of dependency relations ()

No Yes

Fixed order: differentiate trees based on surface word order ()

No Yes

Head: specify potential restrictions on the head node ()

Association measures: print MI, MI3, Dice, logDice, t-score and simple-LL scores ()

No Yes

{% if 'general' in validation %}

{% endif %}

Input data

Tree specification

add Advanced settings

Advanced settings