| Interface | Description |
|---|---|
| Eval |
An interface which can be implemented by anything that evaluates
one tree at a time and then prints out a summary when done.
|
| ParserQueryEval |
Evaluate based on the ParserQuery rather than the Tree produced
|
| Class | Description |
|---|---|
| AbstractEval |
A framework for Set-based precision/recall/F1 evaluation.
|
| AbstractEval.CatErrorEval |
This class counts which categories are over and underproposed in trees.
|
| AbstractEval.RuleErrorEval | |
| AbstractEval.ScoreEval |
This isn't really a kind of AbstractEval: we're sort of cheating here.
|
| BestOfTopKEval |
Applies an AbstractEval to a list of trees to pick the best tree
using F1 measure.
|
| Evalb |
A Java re-implementation of the evalb bracket scoring metric (Collins, 1997) that accepts Unicode input.
|
| Evalb.CBEval | |
| EvalbByCat |
Computes labeled precision and recall (evalb) at the constituent category level.
|
| FilteredEval |
An AbstractEval which doesn't just evaluate all constituents, but
lets you provide a filter to only pay attention to constituents
formed from certain subtrees.
|
| LeafAncestorEval |
Implementation of the Leaf Ancestor metric first described by Sampson and Babarczy (2003) and
later analyzed more completely by Clegg and Shepherd (2005).
|
| TaggingEval |
Computes POS tagging P/R/F1 from guess/gold trees.
|
| TopMatchEval |
Measures accuracy by only considering the very top of the parse tree, eg where S, SINV, etc go
|
| TreeSpanScoring |
Provides a method for deciding how similar two trees are.
|
| UnlabeledAttachmentEval |
Dependency unlabeled attachment score.
|