NILC at ASSIN 2: Exploring multilingual approaches

Marco A.Sobrevilla Cabezudo; Marcio Inácio; Ana Carolina Rodrigues; Edresson Casanova; Rogério Figueredo De Sousa

NILC at ASSIN 2: Exploring multilingual approaches

Marco A.Sobrevilla Cabezudo, Marcio Inácio, Ana Carolina Rodrigues, Edresson Casanova, Rogério Figueredo De Sousa

Universidade de São Paulo

Research output: Contribution to journal › Conference article › peer-review

Abstract

Recognizing Textual Entailment, also known as Natural Language Inference recognition, aims to identify if it is possible to infer the meaning of a text from another fragment of text. In this work, we investigate the use of multilingual models, through BERT, for recognizing inference and similarity in the ASSIN 2 dataset, an entailment recognition and sentence similarity corpus for Portuguese. We also investigate possible features that could enhance the results, such as similarity scores or WordNet relations. Our results show that a multilingual pre-trained BERT model may be su.

Original language	English
Pages (from-to)	48-57
Number of pages	10
Journal	CEUR Workshop Proceedings
Volume	2583
State	Published - 2020
Externally published	Yes
Event	2nd ASSIN Shared Task: Evaluating Semantic Textual Similarity and Textual Entailment in Portuguese, ASSIN 2019 - Salvador, Brazil Duration: 15 Oct 2019 → …

Keywords

BERT
Cross-lingual Training
Multilingual Training
Natural Language Inference

Cite this

@article{8038aeb5a75d4d80817cd575a9c0d8d0,

title = "NILC at ASSIN 2: Exploring multilingual approaches",

abstract = "Recognizing Textual Entailment, also known as Natural Language Inference recognition, aims to identify if it is possible to infer the meaning of a text from another fragment of text. In this work, we investigate the use of multilingual models, through BERT, for recognizing inference and similarity in the ASSIN 2 dataset, an entailment recognition and sentence similarity corpus for Portuguese. We also investigate possible features that could enhance the results, such as similarity scores or WordNet relations. Our results show that a multilingual pre-trained BERT model may be su.",

keywords = "BERT, Cross-lingual Training, Multilingual Training, Natural Language Inference",

author = "Cabezudo, {Marco A.Sobrevilla} and Marcio In{\'a}cio and Rodrigues, {Ana Carolina} and Edresson Casanova and {De Sousa}, {Rog{\'e}rio Figueredo}",

note = "Publisher Copyright: {\textcopyright} 2020 for this paper by its authors.; 2nd ASSIN Shared Task: Evaluating Semantic Textual Similarity and Textual Entailment in Portuguese, ASSIN 2019 ; Conference date: 15-10-2019",

year = "2020",

language = "English",

volume = "2583",

pages = "48--57",

journal = "CEUR Workshop Proceedings",

issn = "1613-0073",

}

TY - JOUR

T1 - NILC at ASSIN 2

T2 - 2nd ASSIN Shared Task: Evaluating Semantic Textual Similarity and Textual Entailment in Portuguese, ASSIN 2019

AU - Cabezudo, Marco A.Sobrevilla

AU - Inácio, Marcio

AU - Rodrigues, Ana Carolina

AU - Casanova, Edresson

AU - De Sousa, Rogério Figueredo

PY - 2020

Y1 - 2020

N2 - Recognizing Textual Entailment, also known as Natural Language Inference recognition, aims to identify if it is possible to infer the meaning of a text from another fragment of text. In this work, we investigate the use of multilingual models, through BERT, for recognizing inference and similarity in the ASSIN 2 dataset, an entailment recognition and sentence similarity corpus for Portuguese. We also investigate possible features that could enhance the results, such as similarity scores or WordNet relations. Our results show that a multilingual pre-trained BERT model may be su.

AB - Recognizing Textual Entailment, also known as Natural Language Inference recognition, aims to identify if it is possible to infer the meaning of a text from another fragment of text. In this work, we investigate the use of multilingual models, through BERT, for recognizing inference and similarity in the ASSIN 2 dataset, an entailment recognition and sentence similarity corpus for Portuguese. We also investigate possible features that could enhance the results, such as similarity scores or WordNet relations. Our results show that a multilingual pre-trained BERT model may be su.

KW - BERT

KW - Cross-lingual Training

KW - Multilingual Training

KW - Natural Language Inference

UR - http://www.scopus.com/inward/record.url?scp=85082673024&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:85082673024

SN - 1613-0073

VL - 2583

SP - 48

EP - 57

JO - CEUR Workshop Proceedings

JF - CEUR Workshop Proceedings

Y2 - 15 October 2019

ER -

NILC at ASSIN 2: Exploring multilingual approaches

Abstract

Keywords

Other files and links

Fingerprint

Cite this