weitechen / anafora Goto Github PK
View Code? Open in Web Editor NEWAnafora is a web-based raw text annotation tool
Anafora is a web-based raw text annotation tool
When annotator creates a link from an entity, Anafora should prevent it links to itself
I got following error opening an in-progress adjudication task
Uncaught TypeError: Cannot read property 'length' of undefined
at Function.each (jquery-1.8.2.min.js:2)
at PropertyType. (anaforaAdjudicationProject.js:914)
at Function.each (jquery-1.8.2.min.js:2)
at Entity. (anaforaAdjudicationProject.js:912)
at Function.each (jquery-1.8.2.min.js:2)
at AnaforaProject. (anaforaAdjudicationProject.js:897)
at Function.each (jquery-1.8.2.min.js:2)
at AnaforaAdjudicationProject.readFromXMLDOM (anaforaAdjudicationProject.js:896)
at String. (annotate.js:217)
at Function.each (jquery-1.8.2.min.js:2)
It'd be handy to have maybe a "Go to..." option in the menu, where you can type in a particular entity/relation ID and have the interface select it.
When reloading the *inprogress.xml, the schema is empty, but i don't know why.
When document contains CR/LR character, the Anafora doesn't count and display the offset correct.
Remember to move and keep the original task name in the merged document .xml file
Hi, there is a spelling mistake in line 665 on the master branch.
from django.core.exceptions import ImpoperlyConfigured
It should be ImproperlyConfigured instead of ImpoperlyConfigured. I found this error when I tried to configure it on my own computer.
Anafora is already supported on Safari (iOS 6) and Chrome (Android, iOS), but lacks ability to click-to-highlight, click in tree to create, required for annotation. Tablet mode could be auto-enabled when mobile chrome (for consistency) useragent is found.
Allows view-as-annotator, easy viewing of all annotations
Hey everyone,
Anafora looks like a really powerful tool, but so far I haven't been able to set it up successfully. Is there any installation guide other that this one? Also, is the project still alive?
Make sure that the deploying work is perfect on any Unix based machine. Will try on a Amazon EC2 server
We've gotten requests for a means to view Anafora docs statically and without Internet access on Windows/Mac/Linux, perhaps via a static site. Is this possible?
Improve the error message when the input file (include .setting.xml, schema files, and annotation file) has formatted errors
The “cancel” button on the screen before you select a specific document doesn’t seem to work. If I accidentally select, say, "colon notes" instead of "brain notes," clicking "cancel" doesn't take me back to the previous screen (or do anything at all, actually); I have to close and restart Anafora.
When adjudicating, the merge function did not leave gold annotations from previous passes as gold directly. It still compares two gold annotation
Sometimes when a markable participates in a lot of links, when you click on the markable to view those links, it's not possible to see them all because they go off the screen and there's no way of scrolling up or down.
We should have an "About Anafora" menu with the version number, logo, author names, etc
Currently have this situation with a schema as "directSetGold" and only allows one annotator.
One annotator starts a new task. Then another annotator starts it at virtually the same time -- since the first annotator hasn't saved yet, Anafora doesn't prevent the second one from starting it. It may be helpful if on a save action (one that would lead to creating a file), Anafora double-checks the file system to see if saving would still be compliant with the allowed number of annotators, and warns the user if it's not.
Use different style for cross / within document annotation
If one entity set itself as linking property, it can not been set at first. However, after we click the entity again, itself has been set as its property still. Should ban this activity
Anafora will go through the directory whenever it's just open a new note or select a note
I'm using Anafora to annotate a corpus of files using a custom schema. However, I've noticed that if I begin to annotate a file and save it, but then leave the site and come back to keep working on it later, I'm not able to access the schema. Even though the file is saved, as long as I walk away from the task and come back I am subsequently unable to mark up the text again.
Any work on the file cannot resume until I delete the existing annotation file on the server, which means losing all the work I had done on it so far. As you can imagine, this is incredibly frustrating. I was wondering if this is something inherent to Anafora or if I'm perhaps doing something wrong.
Any help you can provide would be greatly appreciated!
Annotation time is recorded. We need an interface to easily view it.
Relations Adjudication is missing a few final tweaks before it's ready for deployment.
The default value for relations type (e.g. THYME TLINKs "contains") isn't being respected or read in Anafora.
To improve usability for annotators when the focus is on the entity properties the tab key should move through the different properties of the entity.
Allow administrators to view RSS feeds of telling when sets have been completed and by who.
I'm currently writing a plugin for GATE (http://gate.ac.uk) to enable us to read in documents annotated with Anafora. You can find the latest version of the plugin at https://github.com/GateNLP/gateplugin-Format_Anafora
While the XML format Anafora uses is nice and easy to parse I've come across quite a few documents where an annotator has somehow managed to produce an annotation that ends after the end of the document. In many cases this isn't just one or two characters difference (something I could understand if there were issues of encoding etc.) but a difference of 50 characters or more. Currently I simply truncate these annotations to match the document length, and it seems that this results in seeing the same annotations in both Anafora and GATE.
As I haven't checked every annotation on every document though I'm wondering if this is an isolated issues with annotations that end at the end of a document, or if there might be a wider issue with annotation offsets being stored incorrectly.
I've attached a document and annotation file so you can see what I mean. In this instance the document is 817 characters long (I'm assuming it's UTF-8 but wiith no multi-byte characters as it's also 817 bytes long), but the second annotation in the file produced by Anafora spans from offset 656 to 835; in otherwords it goes 18 characters beyond the end of the document.
As you can see in ID 169 here (span "bends"), Anafora is claiming a disagreement in the "part_of" property, when they are actually referring to the same entity ("fault").
If we look at that entity, we can see that this is because the annotators disagreed on the properties of the "fault".
Anafora should mark the latter as a disagreement, but not the former. So perhaps comparing just span indices and entity types of the properties, not the properties of properties.
Hi,
I haven't yet understood how to handle multiple users. Could you briefly describe how to create / add new users?
We are very impressed with Anafora's features. We intend to try your application for annotating custom transcription documents.
We have just managed to install the application, but we have been encountering issues loading content. Could you suggest a way forward?
The goal is to be able to load our document and view annotations that works exactly like the demo page.
Annotators have requested that docs marked as completed be ordered by date completed, rather than by number. Is this feasible?
If you compare these two screenshots… One has just the date highlighted because I highlighted from the beginning to the end. The other has the date plus a little bit of whitespace afterwards because I started highlighting on the line below the date and just dragged up to select the text. Doing it that way also selected the newline character to be included in the span. Once you actually hit a hot key to mark the span, that whitespace character isn’t coloured/highlighted, so you can't see anymore that you've accidentally included some whitespace.
When I annotate some files the tag in the resulting Anafora XML files are always:
<schema path="./" protocol="file">temporal.schema.xml</schema>
when the schema being used is actually my own schema.
Annotators should have the ability, when a set is complete for them and from adjudication, to click an icon and view their annotations in a sort of adjudication mode next to the gold standard for learning.
It would save at least five mouse clicks to open the next document in the current corpus when a document is marked as completed.
When Entity adjudication is completed, preannotations for relations can be automatically created, specified by project/schema. This is currently done by script.
Ability to use instance-of="*" to allow a list slot to be filled by any entity type
The documentation states :
When you double-click on a word, most browsers will expand the selection to include the entire word.
This works on Edge, but on Firefox or Chrome double-clicking on a word selects that word plus the following word, which is not very ergonomic. It would be nice to have a fix on that.
Because TLINK/ALINK are not allowable types for filling the Source/Target slots, all relations disappear from the grid when you create a new relation. This is the expected behavior, but is frustrating. Perhaps always display relations, even when they're not allowed in the slot, because that behavior is confusing annotators and me.
Give adjudicator the ability to restart the adjudication task.
I have a lot of annotated documents that we did in Anafora 1.0. When I try to open these in Anafora 1.1 the document opens but no schema is loading. Was there backward incompatible change made in the formal of the annotation documents between versions?
Annotator access to different projects in the central Anafora database should be controlled using existing Unix groups.
Anafora creates invalid XML if you use characters like "<" in a text field. For example, if you have a property named "Years" which has a text field for entry, then you get an Anafora XML file containing invalid XML like:
<Years><1</Years>
The comparison of coreference chain doesn't work well. Check AMR/Analysis
In order to improve usability for the annotators, when a new entity is created the focus should go to the first property of the entity.
The number of total annotation and completed annotation are wrong.
I don't have more details than this about the circumstances, I'm afraid, but it's been reported that sometimes when an annotator goes to mark a note as completed, they find the button in the menu is greyed out. But apparently if they click it, the note will get marked as completed appropriately.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.