asreview/asreview

Active learning for systematic reviews

What it solves

ASReview LAB accelerates the process of screening large textual datasets for systematic reviews and meta-analyses. It eliminates the need to manually screen every single record by using AI to prioritize the most relevant documents, reducing the time and effort required to find relevant papers.

How it works

The tool uses an active learning loop: the user labels a small set of records as "Relevant" or "Not Relevant," and the AI model continuously learns from these decisions to re-rank and prioritize the most likely relevant records for the user to review next. Users can provide prior knowledge to jumpstart the process and monitor progress via a dashboard.

Who it’s for

Researchers, academics, and professionals conducting systematic reviews or any project requiring the systematic screening of large volumes of text data.

Highlights

  • Active Learning: AI models that adapt in real-time to user labeling decisions.
  • Scientifically Validated: Methodology published in Nature Machine Intelligence.
  • Flexible AI Models: Support for pre-configured ELAS models or custom components.
  • Privacy First: Open-source software that collects no user or usage data.
  • Duplicate Management: Automatic hiding of duplicate records to streamline the workflow.
  • Simulation Toolkit: Tools to assess model performance on already labeled datasets.

Related

  • Project
    HumanSignal/label-studio

    An open-source data labeling tool that allows users to annotate audio, text, images, and video to prepare or improve training data for machine learning models.

  • Project
    CVHub520/X-AnyLabeling

    A unified cross-platform desktop application for AI-assisted annotation of text, image, video, and multimodal data, integrating state-of-the-art deep learning models to automate labeling.

  • Project
    Future-Scholars/paperlib

    An open-source academic paper management tool that simplifies metadata scraping and organization, featuring LLM-powered summarization and semantic search.

  • Project
    NeuroDong/Ai-Review

    An AI-powered paper review system that provides structured feedback, visual layout analysis, and competitiveness ranking against other manuscripts.

  • Project
    LAION-AI/CLIP_benchmark

    A standardized evaluation framework for CLIP-like models to measure performance on zero-shot classification, retrieval, captioning, and linear probing across diverse datasets.