arborator/arborator-server

The Arborator software is aimed at collaboratively annotating dependency corpora.

arborator

The Arborator software is aimed at collaboratively annotating dependency corpora.

Please simply try it out on the access to the Arborator (no login is needed)!

Click on the Help link on top to understand the features.

The quick page allows to easily

• visualize and edit a CoNLL file of dependency annotation
• create a dependency graph for a yet unannotated sentence
• export in graphical or CoNLL formats.

A report on Arborator's options for class-sourcing dependency annotation is here: https://scholar.google.fr/citations?view_op=view_citation&hl=en&citation_for_view=jCFpJfEAAAAJ:k_IJM867U9cC

Please also cite this article if you use Arborator for research.

Check out the instructions on the Wiki page.

arborator-server's Issues

Publicize on UD website?

Is there a blurb/screenshot I can post to http://universaldependencies.org/tools.html?

Twitter API key preventing PR?

I'm having trouble becoming even with arborator-server/dev because I can't commit this file:

arborator-server/config.py

This means that PRs from my fork to fix #19 are getting conflicts. A GitHub bot is e-mailing me saying "GitGuardian detected an API key from Twitter". It appears the file does contain a Twitter API key, not sure whose, so it's probably a good idea to clean this up on origin dev (and shut down the API key).

@kimgerdes can you remove this file from dev if it's not needed for anything? I can also PR into stable if you like, that doesn't conflict. I need to edit project.cgi to resolve #19 .

Question: editing project properties

Hi, I just upgraded to the latest master and noticed that inside a project, the thing black bar at the bottom where we could enter project configuration and e.g. edit relations/labels available in the project is gone.

Does anyone know how we can access this functionality in the latest version?

XSS vulnerability

The project parameter is not properly escaped in project.cgi, leading to a possible XSS vulnerability, e.g.:

<SERVER>/project.cgi?project=orpheo"><script>alert(150)</script>

Mouse drag-and-drop doesn't work normally for laptops with touch screens

Several of my students encountered this with Georgetown's installation of Arborator. They had to resort to a workaround of touching at the starting point and the ending point to form the edge. One student suggested that it may be due to an outdated version of the raphael library. (I myself don't have a laptop with a touch screen.)

issue with mixing files

While I was uploading files (to: https://arboratorgrew.elizia.net/#/) in a project, afterwards files annotated by
different annotators were mixed and merged under one filename and
additionally the order of the sentences was re-arranged(!) 😱

any idea of what could have caused this issue?

Issue with export function

I noticed an issue with the conll export function in the last commit of Arborator (17-10-08). After updating my installation to this last commit, I was unable to export conll trees properly; conll files were exported but the exported files in the export folder lacked dependency information. All other fields where correctly exported, included the conll comments.

The problem appears to be with the definition of the function conllSentenceExport in database.py. Using the function definition from before the last commit allowed me to export the files with the dependency information. However, the old version does not export the conll comments.

In addition, while exporting individual trees (e.g. in pdf or svg) from the editor window, I encountered this error:

convert_svg.cgi.pdf

This is solved by creating a "cache" folder in arborator/export. I am not sure if I missed something during the installation of arborator or if this folder should be included in the source dump.

Arborator displays are clipped

While Arborator has a very slick UI it seems it is not working properly. I have the latest Firefox Quantum v 60 and I use it on a MAC.

From time to time the displayed node graph distorts and we see this:

cannot completely delete users

User names still appear in "assign" drop-down menu from the main interface after deleting them using the user administration window, and checking that there is no reference to these users in any file of the "users" folder. Is this an issue with the update to the database after the delete user command is issued?

invalid http headers?

I recently attempted to reinstall arborator (commit a16f3a3) on a new server which runs Ubuntu 16.04 and apache2 (2.4.18). When trying to access arborator in firefox, I get a 500 internal server error. I can run other CGI scripts from the directory where arborator is installed (/var/www/html/), so the problem is specific to my arborator installation. Apache2's error log reads:

[http:error] ... AH02429: Response header name '<!--' contains invalid characters ....

These posts suggest there might be an issue with a cgi script that outputs an invalid http header:

https://serverfault.com/questions/853103/500-internal-server-ah02429-response-header-name
https://blog.tigertech.net/posts/apache-cve-2016-8743/

I haven't been able to find the issue in arborator's cgi scripts. I wonder if there is indeed an issue with these scripts or if my server is not properly configured.

Splitting a sentence

The wiki says

• splitting the sentence at any position into two separate lines. To do that, the administrator selects the word after which he or she wants to split the sentence. The word is then color highlighted. Then, a Ctrl-click on the chain button, opens an approval dialog.

After I click on a word, Ctrl-clicking the chain button gives the same behavior as regular clicking—a dialog pops up confirming that I want to join the sentences. Is splitting still supported? (I'm using a Mac, where Cmd is the equivalent of Ctrl on Windows—could that be a problem?)

no createDB.py file? How to create project database

There is no createDB.py file as described in the wiki. How should we create it for a project?

Python 2.6 incompatible syntax in lib/treebankfiles.py

The double with statement here crashes on Python 2.6:

https://github.com/Arborator/arborator-server/blob/stable/lib/treebankfiles.py#L1303

This is important for servers running RHEL, which comes with Python 2.6 by default.

Add support for .conllu format?

Thanks for an excellent tool! I've started using it for annotating with Universal Dependencies, which normally use the CoNLL-U format. This is an enhancement to CoNLL10, the main difference being the ability to represent multiword tokens. I have been stripping out the multiword lines before uploading to Arborator, but it would be nice if they were preserved, and even better if they were displayed (e.g., by underlining groups of tokens). How hard would this be to add?

Is there a way to mark a node as the root of the sentence?

I've just been leaving the root node unattached, but in the exported CoNLL, it is then _ rather than 0.

About admin account

I have installed arborator-server and the arborator start page, but I can't log in with admi account. Click the login button without any hints.Can you help me?

2nd document export option only exports 1 tree

On the Georgetown Arborator installation, for a document, "export all trees by any annotator as CoNLL file" stops prematurely. E.g., for one of the documents, it only writes 1 tree to the file even though there are 5 annotations. Any idea why this could be happening? @amir-zeldes and I haven't had a chance to sit down and try to debug it.

The first option, "export complete files per assignment", works but makes it difficult to tell in the exported data which trees correspond to saved annotations and which are defaults from the parser. Annotators were asked to save a copy of the parser output if it was correct, but in the export there is no way to distinguish sentences they saved with no changes vs. sentences they skipped.

Documenting advanced functionality

Moving from #3:

Exo (exercise) modes:

• teacher visible mode: dumb exercise where students have to copy the teacher's tree which is visible but not directly modifiable. a good start with 3 sentences or so.
• no feedback: the student can't see the teacher's tree and gets no feedback, but the admin can export the results of the students' annotations compared to the teacher trees
• percentage: when students save, they can see how many percent they got wrong of dependencies and pos, but they don't know where
• graphical feedback: when students save, they can see where there are problems compared to the teacher's tree and they have to find the right annotation.

baseAnnotatorName = parser

baseAnnotatorName is no longer used. to make it clearer that this is important only for exercise modes the variable is called exoBaseAnnotatorName. it is the name of the user that provides the tree that the student should start with.
teacher is used for comparisons and results in exercises.

I assigned the entire sample to the user. Is there a way to assign only particular sentences?

yes, but not from the graphical interface. there's code to intelligently distribute new sentences mixed up with already annotated sentences in a way that two students don't have the same set of sentences to annotate. this is used to class-source dependency annotation...

Issue with project overview link

When clicking on the project overview link at the top left corner of the editor window ("XYZ annotation project"), from within an annotation window, arborator frequently sends me back to the project login screen, instead of the project overview screen. This is not systematic though. I can't reproduce the bug systematically, but it happens virtually every time I use arborator.

This is a recent issue. I'm using a fresh install from a version of arborator downloaded on June 23, 2020, with Firefox 77.0.1 (64-bit) on linux Mint 19.1.

For some annotations, all tokens show up in annotation view as "undefined"

No dependency arcs are showing up (it's unclear if the user has been able to annotate any). The POSes are intact, though. And other users' annotations for the sentence show up fine.

CoNLL export suppresses tag2

The second POS tag column is exported in XML export, but is always _ in conll export.