Annotation platform

Table of Contents

1. About The Project
2. Implementation
3. Prerequisites
4. Dependencies
5. Authors

About The Project

This project is about an annotation platform that enables users to annotate data by uploading a dataset. It offers the possibility to set the desired value of the inter-annotator kappa agreement coefficient, which determines the level of agreement required among annotators to annotate unlabeled data from the dataset. This customization ensures the production of high-quality annotated data. Additionally, the platform allows administrators to set the number of documents each user should annotate once they have passed the defined threshold.

The methodology is described in the article:

Molina-Villegas, A., Cattin T., Gazca-Hernandez K. & Aldana-Bobadilla E. (2023). High-quality Data from Crowdsourcing Towards the Creation of The Mexican Anti-immigrant Speech Corpus. Applied Sciences. Section: Computing and Artificial Intelligence. ISSN 2076-3417 (in press)

If you want a copy of the related data please send us an email:

karina.gazca@cinvestav.mx, edwyn.aldana@cinvestav.mx,  amolina@centrogeo.edu.mx

(back to top)

Implementation

Figure 1. Admin panel of the annotation platform.

Figure 2. User interface for annotating tweets in the platform.

(back to top)

Prerequisites

Python 3.X installed

python --version

Python 3.10.0

(back to top)

Dependencies

pandas==1.3.5
cryptography==38.0.1
Flask==2.0.2
Flask_Session==0.4.0
Flask_SQLAlchemy==2.5.1
numpy==1.23.0
scikit_learn==1.2.2
SQLAlchemy==1.4.36
Werkzeug==2.0.2

(back to top)

Authors

Karina Yaneth Gazca Hernández

Alejandro Molina Villegas

Thomas Cattin

Edwyn Javier Aldana Bobadilla

(back to top)