ActiveTigger

Key Features

Everything you need to annotate, explore, and classify textual data.

✍

Collaborative Annotation

Collaboratively annotate text data with multiple users, stabilise codebooks, and manage annotation campaigns.

⚡

Active Learning

Reduce annotation effort by letting the model suggest the most informative samples to label next.

🤖

BERT Fine-Tuning

Fine-tune encoder classifiers directly from the interface to build high-performance text classifiers.

📊

Topic Exploration

Explore your corpus with BERTopic to discover themes and structure in your data.

💬

Generative AI

Leverage generative models via API calls to assist or automate annotation at scale.

📈

Evaluation & Export

Evaluate classifier performance, compare annotators, and export results for further analysis.

Our Philosophy

Guiding principles behind the project.

Facilitate collaborative human/AI annotation and exploration of textual data.
Democratize access to computational social science tools for diverse communities.
Build a lightweight, frugal, and open-source application.

Repositories

The ActiveTigger ecosystem on GitHub.

TypeScript / Python

The main application — text annotation web tool dedicated to computational social sciences.

at-cli-client

Python

Python wrapper to use the ActiveTigger API programmatically from scripts or notebooks.

at-data

Data

Sample datasets and data resources for use with ActiveTigger.

documentation

Docs

User and developer documentation — quickstart guides, tutorials, and reference.

The Team

ActiveTigger is developed in the CREST research unit (CNRS, Polytechnique, ENSAE), within the CSS@IP-Paris team, with the help of OuestWare.

The current version is a refactor of the original R Shiny ActiveTigger app by Julien Boelaert & Etienne Ollion. The name "Active Tigger" is a pun on the similarity between "Tagger" and "Tigger."

Generously funded by DRARI Île-de-France, Progedo, and CREST.