umu.sePublications
Change search
ReferencesLink to record
Permanent link

Direct link
EM-training for probabilistic aligned hypergraph bimorphisms
Umeå University, Faculty of Science and Technology, Department of Computing Science. (Foundations of Language Processing)ORCID iD: 0000-0001-7349-7693
Technische Universität Dresden.
Technische Universität Dresden.
2016 (English)In: Proc. StatFSM 2016: ACL Workshop on statistical NLP and weighted automata, The Association for Computational Linguistics , 2016Conference paper (Refereed)
Abstract [en]

We define the concept of probabilistic aligned hypergraph bimorphism. Each such bimorphism consists of a probabilistic regular tree grammar, two hypergraph algebras in which the generated trees are interpreted, and a family of alignments between the two interpretations. It generates a set of bihypergraphs each consisting of two hypergraphs and an alignment between them; for instance, discontinuous phrase structures and non-projective dependency structures are bihypergraphs. We show an EM-training algorithm which takes a corpus of bihypergraphs and an aligned hypergraph bimorphism as input and calculates a probability assignment to the rules of the regular tree grammar such that in the limit the maximum-likelihood of the corpus is approximated.

Place, publisher, year, edition, pages
The Association for Computational Linguistics , 2016.
Keyword [en]
EM training, bimorphism, hyperedge replacement, maximum likelihood
National Category
Computer Science Language Technology (Computational Linguistics)
Research subject
Computer Science; datorlingvistik
Identifiers
URN: urn:nbn:se:umu:diva-121676OAI: oai:DiVA.org:umu-121676DiVA: diva2:933575
Conference
StatFSM 2016: ACL Workshop on statistical NLP and weighted automata
Available from: 2016-06-06 Created: 2016-06-06 Last updated: 2016-06-06

Open Access in DiVA

No full text

Search in DiVA

By author/editor
Drewes, Frank
By organisation
Department of Computing Science
Computer ScienceLanguage Technology (Computational Linguistics)

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 54 hits
ReferencesLink to record
Permanent link

Direct link