public class QuestionClassificationIncrementalLearning
extends Object
This class shows how to use Kelp to build a Question classifier. In the
example, several kernels can be used to build your own classifier.
The following kernels have been implemented:
- bow: a Linear Kernel applied to a boolean Bag-of-Word vector, where each
boolean dimension indicates the presence of the corresponding word in the
question.
- stk: a Subset Tree Kernel [Moschitti, 2006] over the Grammatical Relation
Centered Tree (GRCT) representation [Croce et al(2011)]
- ptk: a Partial Tree Kernel [Moschitti, 2006] over the Grammatical Relation
Centered Tree (GRCT) representation [Croce et al(2011)]
- sptk: a Smoothed Partial Tree Kernel over the Lexically Centered Tree (LCT)
representation [Croce et al(2011)]
- csptk: Compositionally Smoothed Partial Tree Kernel over the Compositional
Lexically Centered Tree (CLCT) representation [Annesi et al(2014)]
A description of the tree representations can be found in [Croce et al(2011)]
and [Annesi et al(2014)].
References:
- [Moschitti, 2006] Alessandro Moschitti. Efficient convolution kernels for
dependency and constituent syntactic trees. In proceeding of European
Conference on Machine Learning (ECML) (2006)
- [Croce et al(2011)] Croce D., Moschitti A., Basili R. (2011) Structured
lexical similarity via convolution kernels on dependency trees. In:
Proceedings of EMNLP, Edinburgh, Scotland.
- [Annesi et al(2014)] Paolo Annesi, Danilo Croce, and Roberto Basili. 2014.
Semantic compositionality in tree kernels. In Proc. of CIKM 2014, pages
1029–1038, New York, NY, USA. ACM
- Author:
- Danilo Croce, Simone Filice, Giuseppe Castellucci