1435 research outputs found
Sort by
A close look at auxiliary problem principles for equilibria
No full textThe auxiliary problem principle allows solving a given equilibrium problem (EP) through an equivalent auxiliary problem with better properties. The paper investigates two families of auxiliary EPs: the classical auxiliary problems, in which a regularizing term is added to the equilibrium bifunction, and the regularized Minty EPs. The conditions that ensure the equivalence of a given EP with each of these auxiliary problems are investigated. This analysis leads to extending some known results for variational inequalities and linear EPs to the general case; moreover, new results are obtained as well. In particular, both new results on the existence and uniqueness of solutions and new error bounds based on gap functions with good convexity properties are obtained under weak quasimonotonicity or weak concavity assumptions
CLaSSES: a new digital resource for Latin epigraphy
Full text linkSUMMARY. CLaSSES (Corpus for Latin Socio-linguistic Studies on Epigraphic textS) is an
annotated corpus for quantitative and qualita-tive sociolinguistic analyses on Latin inscrip-tions. It allows specific researches on phono-logical and morphophonological phenomena
of non-standard Latin forms with crucial ref-erence to the typology of the text, its origin
and chronological collocation. This paper
presents the first macrosection of CLaSSES,
focused on the inscriptions from the archaic-early period.
RIASSUNTO. CLaSSES (Corpus for Latin Socio-linguistic Studies on Epigraphic textS) è un
corpus annotato finalizzato all’analisi socio-linguistica quantitativa e qualitativa delle e-pigrafi latine. Permette di analizzare i feno-meni fonologici e morfofonologici che carat-terizzano le forme latine non standard, in re-lazione alla tipologia testuale, all’area geo-grafica di provenienza e alla datazione delle
iscrizioni. L’articolo presenta la prima ma-crosezione di CLaSSES, incentrata sulle i-scrizioni risalenti al periodo preletterario e
arcaic
Corpus ICoN: una raccolta di elaborati di italiano L2 prodotti in ambito universitario
No full textSUMMARY.
The paper describes the essential features of the Corpus ICoN. The corpus includes essays created over 13 years by university students; the essays are divided into two comparable subcorpora dedicated respectively to students who speak Italian as L1 and those who
know the language as L2/FL.
RIASSUNTO.
Il contributo presenta le caratteristiche essenziali del Corpus ICoN. Il corpus raccoglie elaborati realizzati nell’arco di 13 anni da studenti universitari; gli elaborati sono ripartiti in due sottocorpora equivalenti dedicati rispettivamente agli studenti che conoscono l’italiano come L1 e a quelli che lo conoscono come L2/LS
Experiments in Identification of Italian Temporal Expressions
No full textSUMMARY.
We describe our experiments in participating to the EV
aluation of Events aNd Temporal Information(EVENTI) task, for the EVALITA 2014 evaluation campaign. We used the HeidelTime tagger extended with a wrapper for the Tanl
POS tagger and tokenizer of the Tanl suite. The rules for recognizing Italian temporal expressions were rewritten and extended after the submission, leading to a 10 point increase in F1 over the Italian rules in the HeidelTime distribution.
RIASSUNTO.
Nell’articolo descriviamo gli esperimenti svolti per la nostra partecipazione al task EValuation of Events aNd Temporal Information(EVENTI), nel’ambito della campagna di valutazione EVALITA 2014. Per il riconoscimento e normalizzazione delle espresioni temporali abbiamo utilizzato il tagger HeidelTime, estendendolo con un wrapper per poter utilizzare il POS tagger e il tokenizer della suite di NLP Tanl. Le regole per il riconoscimento delle espressioni temporali in italiano sono state riscritte ed estese, dopo la sottomissione, ottenendo un miglioramento di 10 punti di F1 rispetto alle regole presenti nella distribuzione di HeidelTime
The CoLing Lab system for Sentiment Polarity Classification of tweets
No full textSUMMARY.
This paper describes the CoLing Lab system for the EVALITA 2014 SENTIment POLarity Classification SENTIPOLC) task. Our system is based on a SVM classifier trained on the rich set of lexical, global and twitter-specific features described in these pages. Over all, our system reached a 0.63 weighted F-score on the test set provided by the task organizers.
RIASSUNTO.
Questo contributo descrive il sistema CoLing Lab sviluppato
per il task di SENTIment POLarity Classification (SENTIPOL
C) organizzato nel contesto della campagna EVALITA
2014. Il nostro sistema è basato su un classificatore SVM addestrato sulle feature lessicali, globali e specifiche del canale twitter descritte in queste pagine. Il nostro sistema raggiunge uno score di circa
0.63 nel test set fornito dagli organizzatori del task
SYMPAThy: Towards a comprehensive approach to the extraction of Italian Word Combinations
No full textSUMMARY.
The paper presents SYMPAThy,a new approach to the extraction of WordCombinations. The approach is new in that it combines pattern-based (P-based)and syntax-based S-based) methods in or-der to obtain an integrated and nified view of a lexeme’s combinatory potential.
RIASSUNTO.
L’articolo presenta SYMPA-Thy, un nuovo metodo per l’estrazione di Combinazioni di Parole. L’originalità dell’approccio consiste nel combinare il metodo basato su sequenze di parti del discorso (P-based) e quello basato sulle dipendenze sintattiche (S-based) per ar-rivare a una visione integrata e unitaria del potenziale combinatorio di un lessem
Comparison of the Complior Analyse device with Sphygmocor and Complior SP for pulse wave velocity and central pressure assessment
Full text linkBACKGROUND: The Complior device (Alam Medical, France) was used in epidemiological studies which established pulse wave velocity (PWV) as a cardiovascular risk marker. Central pressure is related, but complementary to PWV and also associated to cardiovascular outcomes. The new Complior Analyse measures both PWV and central blood pressure during the same acquisition. The aim of this study was to compare PWV values from Complior Analyse with the previous Complior SP (PWVcs) and with Sphygmocor (PWVscr; AtCor, Australia), and to compare central systolic pressure from Complior Analyse and Sphygmocor.
METHOD: Peripheral and central pressures and PWV were measured with the three devices in 112 patients. PWV measurements from Complior Analyse were analysed using two foot-detection algorithms (PWVca_it and PWVca_cs). Both radial (ao-SBPscr) and carotid (car-SBPscr) approaches from Sphygmocor were compared to carotid Complior Analyse measurements (car-SBPca). The same distance and same calibrating pressures were used for all devices.
RESULTS: PWVca_it was strongly correlated to PWVscr (R(2) = 0.93, P < 0.001) with a difference of 0.0 ± 0.7 m/s. PWVca_cs was also correlated to PWVcs (R(2) = 0.90, P < 0.001) with a difference of 0.1 ± 0.7 m/s. Central systolic pressures were strongly correlated. The difference between car-SBPca and ao-SBPscr was 3.1 ± 4.2 mmHg (P < 0.001), statistically equivalent to the difference between car-SBPscr and ao-SBPscr (3.9 ± 5.8 mmHg, P < 0.001), whilst the difference between car-SBPca and car-SBPscr was negligible (-0.7 ± 5.6 mmHg, P = NS).
CONCLUSION: The new Complior Analyse device provides equivalent results for PWV and central pressure values to the Sphygmocor and Complior SP. It reaches Association for the Advancement of Medical Instrumentation standard for central blood pressure and grades as excellent for PWV on the Artery Society criteria. It can be interchanged with existing devices
Adapting Linguistic Tools for the Analysis of Italian Medical Records
Full text linkSUMMARY: Weaddress the problem ofrecogni-tion ofmedical entities in clinical records
written in Italian. We report on experiments
performed on medical data in English provid-ed in the shared tasks at CLEF-ER 2013 and
SemEval 2014. This allowed us to refine
Named Entity recognition techniques to deal
with the specifics of medical and clinical lan-guage in particular. We present two ap-proaches for transferring the techniques to Italian. One solution relies onthe creation of an Italian corpus of annotated clinical records
and the other onadapting existing linguistic
tools to the medical domain.
RIASSUNTO. Questo lavoro affronta il problema
del riconoscimento di entità mediche inrefer-ti medici in lingua italiana. Riferiamo su de-gli esperimenti svolti su testi medici in ingle-se forniti nei task di CLEF-ER 2013 e SemE-val 2014. Questi ci hanno consentito di raffi-nare le tecniche di Named Entity recognition
per trattare le specificità del linguaggio me-dico e in particolare quello dei referti clinici.
Presentiamo due approcci al trasferimento di
queste tecniche all’italiano. Una soluzione
consiste nella creazione di un corpus di refer-ti medici in italiano annotato con entità me-diche e l’altro nell’adattarestrumenti tradi-zionali per l’analisi linguistica al dominio
medico
“Il Piave mormorava…”: Recognizing Locations and other Named Entities in Italian Texts on the Great War
No full textSUMMARY.
Increasing amounts of sources about World War I (WWI) are nowadays available in digital form. In this paper, we illustrate the automatic creation of a
NE-annotated domain corpus used to adapt an existing NER to Italian WWI texts. We discuss the annotation of the ntraining and test corpus and provide results of the system evaluation.
RIASSUNTO.
Negli ultimi anni, si sono resi disponibili
in formato digitale un numero sempre maggiore di materiali riguardanti la Prima Guerra Mondiale. In questo
lavoro illustriamo la creazione automatica di un corpus di addestramento per adattare un NER esistente a testi italiani sulla Prima Guerra Mondiale e presentiamo i
risultati della valutazione del nostro sistema addestrato sul nuovo corpus
Dependency Parsing Techniques for Information Extraction
No full textSUMMARY.
Dependency parsing is an important component in information extraction, in particular when using suitable formalisms and accurate and efficient
parsing techniques. We review recent advances in dependency parsing and describe our own contribution in the context of the Evalita 2014 DPIE task.
RIASSUNTO.
Il parsing a dipendenze è un componente importante nell’estrazione di informazione da testi, in particolare quando usato con una rappresentazione
adeguatae tecniche di parsing accurate ed efficienti.
Accenniamo agli sviluppi recenti nel parsing a dipendenze e descriviamo il nostro contributo nel contesto del task DPIE di Evalita 2014