Real world NLP models made easy
Domain experts can quickly start labeling their data through an intuitive user interface.
While domain experts label their data, Label Sleuth automatically trains in the background appropriate machine learning models.
To avoid wasted labeling effort, Label Sleuth employs active learning techniques to guide the user in what they should be labeling next.
"Anyone can take advantage of Label Sleuth to quickly annotate high-quality text datasets at a cost lower than ever before."
Prof. Toby Li, University of Notre Dame
Developed by the research community
Developed by researchers across industry and academia, Label Sleuth incorporates latest research from human computer interaction, natural language processing, and artificial intelligence.Read the publications
Label Sleuth has been designed with an extensible architecture allowing the easy integration of new components, such as additional model architectures or active learning techniques.System architecture
Label Sleuth is an open source project welcoming contributions by the open source community.How can I contribute?
"It is critical for machines to learn in a label efficient manner. Label Sleuth achieves this goal for challenging text classification and NLP tasks using a unique combination of very good user-interface and good backend active learning algorithms."
Prof. Rishabh Iyer, UT Dallas
Identify contract clauses of interest (e.g., clauses related to Warranties)
Identify within a large set of text messages bullying content as it begins in order to stop it
Classify customer interactions across different dimensions of interest (e.g., request types, sentiment, etc.)
Get started in 4 easy steps
Let's make sure you have a separate Python environment for Label SleuthGet Anaconda
Let's setup your Python enviroment
Open a terminal or restart it if already open
conda create --yes -n label-sleuth python=3.9
conda activate label-sleuth
Install the system
pip install label-sleuth
Label Sleuth collaborators include