Multilingual Semantic Relatedness using Lightweight Machine Translation

Siamak Barzegar; Brian Davis; Siegfried  Handschuh; Andre Freitas

doi:10.1109/ICSC.2018.00024

Multilingual Semantic Relatedness using Lightweight Machine Translation

Siamak Barzegar, Brian Davis, Siegfried Handschuh, Andre Freitas

Information Management

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

79 Downloads (Pure)

Abstract

Distributional semantic models are strongly dependent on the size and the quality of the reference corpora, which embeds the commonsense knowledge necessary to build comprehensive models. While high-quality texts containing large-scale commonsense information are present in English, such as Wikipedia, other languages may lack sufficient textual support to build distributional models. This paper proposes using the combination of a lightweight (sloppy) machine translation model and an English Distributional Semantic Model (DSM) to provide higher quality word vectors for languages other than English. Results show that the lightweight MT model introduces significant improvements when compared to language-specific distributional models. Additionally, the lightweight MT outperforms more complex MT methods for the task of word-pair translation.

Original language	English
Title of host publication	2018 IEEE 12th International Conference on Semantic Computing (ICSC)
Publisher	IEEE
Pages	108-114
Number of pages	7
ISBN (Electronic)	78-1-5386-4408-9
ISBN (Print)	978-1-5386-4409-6
DOIs	https://doi.org/10.1109/ICSC.2018.00024
Publication status	Published - 12 Apr 2018

Keywords

Machine Translation
Multilingual Distributional Semantic Models
Semantic Relatedness
Semantic Similarity

Access to Document

10.1109/ICSC.2018.00024

ieee-icsc-semantic-1Accepted author manuscript, 452 KB

Cite this

@inproceedings{e8ed246eaf0543408e570b0b5cebd6ed,

title = "Multilingual Semantic Relatedness using Lightweight Machine Translation",

abstract = "Distributional semantic models are strongly dependent on the size and the quality of the reference corpora, which embeds the commonsense knowledge necessary to build comprehensive models. While high-quality texts containing large-scale commonsense information are present in English, such as Wikipedia, other languages may lack sufficient textual support to build distributional models. This paper proposes using the combination of a lightweight (sloppy) machine translation model and an English Distributional Semantic Model (DSM) to provide higher quality word vectors for languages other than English. Results show that the lightweight MT model introduces significant improvements when compared to language-specific distributional models. Additionally, the lightweight MT outperforms more complex MT methods for the task of word-pair translation.",

keywords = "Machine Translation, Multilingual Distributional Semantic Models, Semantic Relatedness, Semantic Similarity",

author = "Siamak Barzegar and Brian Davis and Siegfried Handschuh and Andre Freitas",

year = "2018",

month = apr,

day = "12",

doi = "10.1109/ICSC.2018.00024",

language = "English",

isbn = "978-1-5386-4409-6",

pages = "108--114",

booktitle = "2018 IEEE 12th International Conference on Semantic Computing (ICSC)",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - Multilingual Semantic Relatedness using Lightweight Machine Translation

AU - Barzegar, Siamak

AU - Davis, Brian

AU - Handschuh, Siegfried

AU - Freitas, Andre

PY - 2018/4/12

Y1 - 2018/4/12

N2 - Distributional semantic models are strongly dependent on the size and the quality of the reference corpora, which embeds the commonsense knowledge necessary to build comprehensive models. While high-quality texts containing large-scale commonsense information are present in English, such as Wikipedia, other languages may lack sufficient textual support to build distributional models. This paper proposes using the combination of a lightweight (sloppy) machine translation model and an English Distributional Semantic Model (DSM) to provide higher quality word vectors for languages other than English. Results show that the lightweight MT model introduces significant improvements when compared to language-specific distributional models. Additionally, the lightweight MT outperforms more complex MT methods for the task of word-pair translation.

AB - Distributional semantic models are strongly dependent on the size and the quality of the reference corpora, which embeds the commonsense knowledge necessary to build comprehensive models. While high-quality texts containing large-scale commonsense information are present in English, such as Wikipedia, other languages may lack sufficient textual support to build distributional models. This paper proposes using the combination of a lightweight (sloppy) machine translation model and an English Distributional Semantic Model (DSM) to provide higher quality word vectors for languages other than English. Results show that the lightweight MT model introduces significant improvements when compared to language-specific distributional models. Additionally, the lightweight MT outperforms more complex MT methods for the task of word-pair translation.

KW - Machine Translation

KW - Multilingual Distributional Semantic Models

KW - Semantic Relatedness

KW - Semantic Similarity

UR - http://www.scopus.com/inward/record.url?scp=85048453030&partnerID=8YFLogxK

U2 - 10.1109/ICSC.2018.00024

DO - 10.1109/ICSC.2018.00024

M3 - Conference contribution

SN - 978-1-5386-4409-6

SP - 108

EP - 114

BT - 2018 IEEE 12th International Conference on Semantic Computing (ICSC)

PB - IEEE

ER -

Multilingual Semantic Relatedness using Lightweight Machine Translation

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this