ML Assisted Labeling

Automatically give label suggestion on your data

Overview

Datasaur’s ML-assisted labeling extension enhances efficiency and accuracy in data labeling for NLP projects. It integrates open-source models, large language models (LLMs), and custom models, providing automatic labeling for both span-based and row-based tasks. This tool streamlines the data labeling process, automating your workflow to save time and improve data quality, allowing you to focus solely on reviewing.

Use Case

Finding out the sentiment of user reviews (Row-based)

You can use ML-assisted labeling features in many ways, one of the core ways you can use is sentiment analysis of reviews. Let’s run through a step-by-step guide on how this can be done.

1. Create a project: Follow the guide here to create a row labeling project.

2. Enable ML-assisted labeling: Click the gear icon from the extension panel on the right to open the Manage extensions dialog, and enable the ML-assisted labeling feature.

   

   Manage Extensions Pop Up
3. Select Sentiment Analysis: Click Sentiment Analysis for the service provider and choose the Target text and Target question.

   

   ML Assisted Labeling Settings - Sentiment Analysis
4. Predict and review labels: Click Predict labels. After processing, review the labels and click Accept or Reject. Voila! Your sentiment analysis is complete.

   

   Prediction Result

Using spaCy to automatically label entire text data (Span-based)

You can also use service provider, spaCy, in ML-assisted labeling feature to automatically label your data. Here’s how you can achieve this:

1. Create project: Just like before, follow the guide here to create a span-based project. Here’s what the data looks like below.

   

   Span Based Project
2. Enable ML-assisted labeling: Open the Manage extensions dialog and turn on the ML-assisted labeling extension.

   

   Manage Extensions Pop Up
3. Select spaCy: Click spaCy for the service provider. To learn more about spaCy click here.

   

   ML Assisted Labeling Settings - spaCy
4. Predict and review labels: Click Predict labels. After processing, review the labels and click Accept or Reject for individual labels. Voila! The labels are applied.

   

   Prediction Result

For further details, please visit the Assisted Labeling - ML Assisted Labeling.