2026-09-13T18:20:26Zhttps://iris.cnr.it/oai/requestoai:iris.cnr.it:20.500.14243/4123652024-05-12T00:21:36Zcom_20.500.14243_68com_20.500.14243_21col_20.500.14243_98ou_ou239
D3. 8 Lexical-semantic analytics for NLP
Federico Martelli
Marco Maru
Cesare Campagnano
Roberto Navigli
Paola Velardi
RafaelJ UreñaRuiz
Francesca Frontini
Valeria Quochi
Jelena Kallas
Kristina Koppel
Margit Langemets
Jesse de Does
Rob Tempelaars
Carole Tiberius
Rute Costa
Ana Salgado
Simon Krek
Jaka ibej
Kaja Dobrovoljc
Polona Gantar
Tina Munda
Martelli, Federico
Maru, Marco
Campagnano, Cesare
Navigli, Roberto
Velardi, Paola
Ureñaruiz, Rafaelj
Frontini, Francesca
Quochi, Valeria
Kallas, Jelena
Koppel, Kristina
Langemets, Margit
de Does, Jesse
Tempelaars, Rob
Tiberius, Carole
Costa, Rute
Salgado, Ana
Krek, Simon
Ibej, Jaka
Dobrovoljc, Kaja
Gantar, Polona
Munda, Tina
research infrastructures
lexicography
lexical resources
word-sense disambiguation
WSD
sense-annotated language data
multilinguality
The present document illustrates the work carried out in task 3.3 (work package 3) focused on lexicalsemantic analytics for Natural Language Processing (NLP). This task aims at computing analytics for lexicalsemantic information such as words, senses and domains in the available resources, investigating their role in NLP applications. Specifically, this task concentrates on three research directions, namely i) which grouping senses based on their semantic similari sense clustering , in ty improves the performance of NLP tasks such as Word Sense Disambiguation (WSD), ii) domain labeling of text , in which the lexicographic resources made available by the ELEXIS project for research purposes allow better performances to be achieved, and fin senses ally iii) analysing the , for which a software package is made available. diachronic distribution of In this deliverable, we illustrate the research activities aimed at achieving the aforementioned goals and put forward suggestions for future works. Importantly, we stress the crucial role played by highquality lexicalsemantic r esources when investigating such linguistic aspects and their impact on NLP applications. To this end, as an additional contribution, we address the paucity of manually the ELEXIS parallelannotated data in the lexical senseannotated datasetsemantic research field and introduce , a novel entirely manuallyavailable in 10 European languages and featuring 5 annotation layers.
2022
info:eu-repo/semantics/other
https://hdl.handle.net/20.500.14243/412365
https://elex.is/wp-content/uploads/ELEXIS_D3_8_Lexical-Semantic_Analytics_for_NLP_final_report.pdf
eng
numberofpages:67
info:eu-repo/grantAgreement/EC/H2020/European Lexicographic Infrastructure