This repository contains the codebase for a span-based text annotation interface. The interface is designed for small-scale projects, with a minimalistic annotation workflow.
To have the interface up and running, you need to install the necessary dependencies.
You can perform these steps on a server or your local machine, depending on where you want to host the interface.
- Clone this repository in a dedicated folder
$ makedir myapp
$ cd myapp
$ gh repo clone cltl/span-annotation-tool
- Assuming you have python3 installed on your machine, do:
$ pip3 install Flask
$ pip3 install Flask-Session
$ pip3 install Flask-Caching
$ pip3 install pandas
-
Put your own annotation guidelines (Annotation_guidelines.pdf) in
./scripts/static. Put your data in the folder./input: this file must be in .tsv format, and have a column with an identificator for each text (e.g., a number). -
Change the
config.inifile according to your needs:
| [general] | |
|---|---|
| title | This will be shown as the annotation title. |
| guidelines_location | Write here the name of the file containing your annotation guidelines, for the users. |
| authorized_users | This must be a comma-separated list of the emails of users who are authorized to take the annotation. |
| [data] | |
|---|---|
| data_file | The name of your annotation input file. |
| text_column | The name of the column with the texts to be annotated. |
| text_id_column | The name of the column containing an id for each text. |
| [annotations] | |
|---|---|
| TAGS_SHORT | Abbreviations for the used tags. If more than one, they must be separated by a -. |
| TAGS_LONG | Plain names of the tags (this is what the annotation users will see), separated by a -. |
| TAG_COLORS | Colors (separated by a -) that will be used to highlight the spans corresponding to the tags. |
| [questions] | |
|---|---|
| span_question | This is the question shown to the users to perform the span annotation task. |
| trigger | The tag abbreviation that triggers the type-in question. If the type-in question is unnecessary, set to None. |
| type_in_question | This question is asked to have the annotators fill in a text input field. As opposed to span_question, this one is optional. If you don't need it, set to None. |
- Run the app
$ python3 app.py
-
Open a browser (if running local, open your browser and go to http://127.0.0.1:5000/ or http://localhost:5000) and annotate!
The result will be stored in
./outputs/is .json format.
This tool is developed as part of the CLARIAH consortium.
