THIVLVC: Retrieval Augmented Dependency Parsing for Latin

Name: THIVLVC: Retrieval Augmented Dependency Parsing for Latin
Uploaded: 2026-09-05T02:12:06.925162+00:00
Duration: 502 s
Description: The episode discusses 'THIVLVC: Retrieval Augmented Dependency Parsing for Latin,' a system that uses a two-stage pipeline combining structural information retrieval and LLM reasoning. Hosts analyze how this approach improves parsing accuracy, particularly in poetry, and conclude that future linguistic AI must embrace annotation ambiguity rather than seeking a single ground truth.

The episode discusses 'THIVLVC: Retrieval Augmented Dependency Parsing for Latin,' a system that uses a two-stage pipeline combining structural information retrieval and LLM reasoning. Hosts analyze how this approach improves parsing accuracy, particularly in poetry, and conclude that future linguistic AI must embrace annotation ambiguity rather than seeking a single ground truth.

Open the full episode