Splitting Documents at Lower Cost: Multi-Split Boundary Decisions for LLM-Based Page Stream Segmentation
It is shown that multi-boundary prediction can make zero-shot page stream segmentation more efficient when the window size is selected for the target corpus, and provided the strongest overall accuracy--efficiency trade-off.
Nikhil Reddy Pottanigari, Sepideh Kharaghani, Saverio Vadacchino et al.
· 0 citations