Conllup JS

Javascript library for converting conll sentence to json format and vice-versa. The code include some test and is written in TS !

Who need that ?

Linguist, NLP researcher/engineer and computer scientist who want to show dependency tree in the browser. It is currently being used in :

• Arborator-Grew : A web-app for online collaborative dependency parsing.
• Reactive-Dep-Tree : An html plugin made with Vue.js for showing interactive dependency tree in the browser
• Surface Syntactic SUD : A guideline on Surface Syntactic Universal Dependencies (SUD). Just check the website and try to interact with the dependency trees to have a live demo :D.

Updates history

3.1.3

• minor change: getNodeFromTreeJson() (previously implemented) throw error in case of non existance of token only if specified

3.1.2

• add: getNodeFromTreeJson() function that returns a node from a tree given it's index that can be a enhanced, group or normal token (indices have different format)

3.1.1

• add: returnTokensInOrder() function that returns tokens of a tree in correct (UD) order (according to group, enhanced and normal tokens rules)

3.1

• add support to Enhanced UD specifications (link here)

3.0

• Change most of variables name to better suit with conllup-py porting
• modify _featuresConllToJson() so it order features in the order specified by UD (thank you @bguil):

In sorting, uppercase letters are considered identical to their lowercase counterparts.

2.1

• add : smart behavior for replaceArrayOfTokens() when a token split is detected. New tokens get their feats replicated from original splitted token, and preserve incoming governors relations

2.0.3

• fix : when tokens are replaced/deleted with replaceArrayOfTokens() method, the tokens pointing to these tokens only got the HEAD deleted but not the DEPREL. It now delet both of them so the outputted conllu is valid

2.0.1

add compatibility for node import (from 3rd party codebase)

2.0.0

(Still in development)

• Refactor the SentenceJson type to move the Group tokens in their own property (SentenceJson.GroupsToken)
• Add support of UDPipe SpacesAfter (=//n ; //t ; //v, //f, //r, /s) MISC logic

1.1.7

• add better error information when importing conll with less or more than 10 columns

1.1.6

• add public constructTextFromTreeJson method (that take into account the SpaceAfter=No misc)

1.1.4

• Fix aggressive normalization of hyphens in FORM and LEMMA

1.1.0

• Support for group token

1-2 it's  _ _ _ _ _ _ _ _
1 it  _ _ _ _ _ _ _ _
2 's  _ _ _ _ _ _ _ _