That's So FETCH: Fashioning Ensemble Techniques for LLM Classification in Civil Legal Intake and Referral
2025-09-10
Summary
The article introduces FETCH, a new ensemble classification method that combines large language models (LLMs) with machine learning (ML) techniques to improve the accuracy of legal issue classification in civil legal intake and referral systems. FETCH demonstrates a classification accuracy of 97.37% using a mix of cost-effective models, surpassing the performance of the current state-of-the-art GPT-5 model.
Why This Matters
Accurate legal issue classification is crucial in ensuring individuals seeking legal help are matched with the appropriate resources, preventing severe consequences like loss of housing or custody. The FETCH method not only enhances accuracy but also significantly reduces costs, making high-quality legal triage more accessible to underfunded legal aid programs and nonprofit organizations.
How You Can Use This Info
Legal professionals and organizations can implement FETCH to improve the efficiency and accuracy of legal referrals, especially in high-stakes situations. By adopting affordable ensemble models, legal aid services can better allocate resources, reduce wait times, and ensure clients receive timely and appropriate legal assistance, ultimately enhancing access to justice.