TiAb Review Plugin: A Browser-Based Tool for AI-Assisted Title and Abstract Screening

Researchers developed TiAb Review Plugin, an open-source Chrome extension that enables AI-assisted screening of academic titles and abstracts without requiring server subscriptions or coding skills. The tool combines Google Sheets for collaboration, Google's Gemini API for LLM-based screening, and an in-browser machine learning algorithm achieving 94-100% recall, demonstrating practical viability for systematic literature reviews.

TiAb Review Plugin addresses a significant accessibility gap in academic research infrastructure by democratizing AI-assisted document screening. Traditional server-based tools create barriers through subscription costs and vendor lock-in, while open-source alternatives typically demand programming expertise. This extension bridges that divide by delivering institutional-grade functionality through a no-code interface that researchers can deploy immediately.

The technical implementation reveals thoughtful architecture. By leveraging Google Sheets as a distributed database rather than maintaining dedicated servers, the tool eliminates infrastructure overhead while enabling multi-reviewer collaboration—critical for systematic reviews requiring consensus. The decision to implement ASReview's active learning algorithm in TypeScript rather than Python demonstrates engineering sophistication; achieving 100% ranking equivalence across validation datasets confirms the port maintained algorithmic fidelity while enabling in-browser execution.

The LLM evaluation results merit scrutiny. The reported 94-100% recall at low precision (2-15%) represents a sensitivity-optimized configuration appropriate for screening stages where missing relevant papers proves costlier than reviewing false positives. Work Saved over Sampling values of 48.7-87.3% at 95% recall indicate substantial efficiency gains compared to manual screening, though absolute precision metrics suggest researchers still require manual validation of model suggestions.

The release positions open-source AI tooling at an inflection point where browser-based execution becomes viable for computationally non-trivial tasks. For academic institutions and individual researchers, this democratization reduces barriers to systematic review adoption. The extension's GitHub availability ensures community contributions and transparency, though long-term sustainability depends on continued maintenance and adaptation as LLM capabilities evolve.

• →Browser extension eliminates server costs and coding requirements for AI-assisted academic screening, democratizing access to advanced tools.
• →TypeScript implementation of machine learning algorithms achieves perfect equivalence with original Python code across validation datasets.
• →Gemini 3.0 Flash with optimized parameters delivers 94-100% recall, reducing manual review burden significantly for systematic literature reviews.
• →Google Sheets integration enables seamless multi-reviewer collaboration without requiring dedicated infrastructure or backend management.
• →Open-source distribution on Chrome Web Store and GitHub establishes community-driven development model for continued enhancement.