Title
An LTAG Perspective on Categorial Inference
Abstract
This paper describes a system of categorial inference based on insights from Lexicalized Tree Adjoining Grammar (LTAG). LTAG is a tree-rewriting system and therefore deals with structural, not string, adjacency. When looked at from the logic perspective, the nodes of the trees become types as in a categorial grammar, with corresponding deductive connections between parent and daughter nodes. The resulting system is based on a hybrid logic, with one logic for building Partial Proof Trees, and the other for composing the partial proofs. We reexamine the use of structural modalities in categorial grammar from this perspective, concluding that the use of structural modalities can be considerably simplified, or even eliminated in some cases. The generative power of the hybrid logic system is beyond context-free, as we demonstrate with a derivation of the cross-serial dependencies in Dutch. The system also inherits polynomial parsing from LTAG.
Year
DOI
Venue
1998
10.1007/3-540-45738-0_6
LACL
Keywords
Field
DocType
ltag perspective,tree-rewriting system,categorial inference,resulting system,hybrid logic system,partial proof trees,lexicalized tree adjoining grammar,structural modality,logic perspective,hybrid logic,categorial grammar
Adjacency list,Tree-adjoining grammar,Hybrid logic,Inference,Computer science,Algorithm,Mathematical proof,Artificial intelligence,Natural language processing,Categorial grammar,Parsing,Hybrid system
Conference
ISBN
Citations 
PageRank 
3-540-42251-X
0
0.34
References 
Authors
2
3
Name
Order
Citations
PageRank
Aravind K. Joshi13479791.99
Seth Kulick222129.66
Natasha Kurtonina39013.20