• No results found

Joint Training of Dependency Parsing Filters through Latent Support Vector Machines

N/A
N/A
Info
Download
Protected

Academic year: 2020

Share "Joint Training of Dependency Parsing Filters through Latent Support Vector Machines"

Copied!
6
0
0

Loading.... (view fulltext now)

Download now ( 6 Page )

Full text

Loading

Figure

Figure 1: The dotted arc can be filtered by labeling any of theboxed roles as True; i.e., predicting that the head t h e3 is not thehead of any arc, or that the modifier h i s 6attaches elsewhere.Role truth values, derived from the gold-standard tree (in grey),are listed adjacent to the boxes, in parentheses.
Figure 2: N o n e´ d o gw`arc’s threshold¯·Φ (¯ 3w¯A hypothetical example of dynamic threshold-ing, where a weak assertion that´should not be a head`·Φ (¯ N a H3) = 0
Table 1: Ablation analysis of intrinsic filter quality.
Table 2: Parsing with jointly-trained filters outperforms independently-trained filters (R+L), as well as a more complexcascade (R+L+Q)

References

Download now ( PDF - 6 Page - 250.55 KB )

Related documents