Beyond Single Parsers: An Empirical Analysis of Dependency Parse Tree Aggregation
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F26%3AV9X2HZNU" target="_blank" >RIV/00216208:11320/26:V9X2HZNU - isvavai.cz</a>
Result on the web
<a href="http://dx.doi.org/10.1007/978-981-96-8197-6_27" target="_blank" >http://dx.doi.org/10.1007/978-981-96-8197-6_27</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1007/978-981-96-8197-6_27" target="_blank" >10.1007/978-981-96-8197-6_27</a>
Alternative languages
Result language
angličtina
Original language name
Beyond Single Parsers: An Empirical Analysis of Dependency Parse Tree Aggregation
Original language description
Dependency parsing is essential in Natural Language Processing (NLP), but parser performance varies across languages and domains, especially in low-resource settings. While aggregation methods have improved other NLP tasks, their role in dependency parsing remains largely unexplored. This study evaluates three unsupervised aggregation frameworks: Maximum Spanning Tree (MST), Conflict Resolution on Heterogeneous Data (CRH), and a Customized Ising Model (CIM), using 71 Universal Dependency test treebanks covering 49 languages. Results show that the CIM consistently outperforms individual parsers and other aggregation approaches by effectively estimating parser quality. These findings highlight the potential of parse tree aggregation for improving parsing robustness in multilingual and low-resource settings. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
Czech name
—
Czech description
—
Classification
Type
D - Article in proceedings
CEP classification
—
OECD FORD branch
10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
Result continuities
Project
—
Continuities
—
Others
Publication year
2025
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Article name in the collection
Lect. Notes Comput. Sci.
ISBN
978-981-96-8196-9
ISSN
—
e-ISSN
—
Number of pages
13
Pages from-to
362-374
Publisher name
Springer Science and Business Media Deutschland GmbH
Place of publication
—
Event location
Sydney
Event date
Jan 1, 2026
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
—