A Phrase Table without Phrases: Rank Encoding for Better Phrase Table Compression

Marcin Junczys-Dowmunt

A Phrase Table without Phrases: Rank Encoding for Better Phrase Table Compression

Marcin Junczys-Dowmunt

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

Abstract

This paper describes the first steps towards a minimum-size phrase table implementation to be used for phrase-based statistical machine translation. The focus lies on the size reduction of target language data in a phrase table. Rank Encoding (R-Enc), a novel method for the compression of word-aligned target language in phrase tables is presented. Combined with Huffman coding a relative size reduction of 56 percent for target phrase words and alignment data is achieved when compared to bare Huffman coding without R-Enc. In the context of the complete phrase table the size reduction is 22 percent.

Original language	English
Title of host publication	16th Annual Conference of the European Association for Machine Translation (EAMT)
Editors	Mauro Cettolo, Marcello Federico, Lucia Specia, Andy Way
Place of Publication	Trento, Italy
Pages	245-252
Number of pages	8
Publication status	Published - 2012
Event	16th Annual Conference of the European Association for Machine Translation (EAMT) - Fondazione Bruno Kessler (FBK) Center for Scientific and Technological Research, Trento, Italy Duration: 28 May 2012 → 30 May 2012

Conference

Conference	16th Annual Conference of the European Association for Machine Translation (EAMT)
Country/Territory	Italy
City	Trento
Period	28/05/12 → 30/05/12

Access to Document

http://hltshare.fbk.eu/EAMT2012/html/Papers/57.pdf

Cite this

@inproceedings{48546bafd2424c07ad195e96674ff23b,

title = "A Phrase Table without Phrases: Rank Encoding for Better Phrase Table Compression",

abstract = "This paper describes the first steps towards a minimum-size phrase table implementation to be used for phrase-based statistical machine translation. The focus lies on the size reduction of target language data in a phrase table. Rank Encoding (R-Enc), a novel method for the compression of word-aligned target language in phrase tables is presented. Combined with Huffman coding a relative size reduction of 56 percent for target phrase words and alignment data is achieved when compared to bare Huffman coding without R-Enc. In the context of the complete phrase table the size reduction is 22 percent.",

author = "Marcin Junczys-Dowmunt",

year = "2012",

language = "English",

pages = "245--252",

editor = "Mauro Cettolo and Marcello Federico and Lucia Specia and Andy Way",

booktitle = "16th Annual Conference of the European Association for Machine Translation (EAMT)",

note = "16th Annual Conference of the European Association for Machine Translation (EAMT) ; Conference date: 28-05-2012 Through 30-05-2012",

}

Junczys-Dowmunt, M 2012, A Phrase Table without Phrases: Rank Encoding for Better Phrase Table Compression. in M Cettolo, M Federico, L Specia & A Way (eds), 16th Annual Conference of the European Association for Machine Translation (EAMT). Trento, Italy, pp. 245-252, 16th Annual Conference of the European Association for Machine Translation (EAMT), Trento, Italy, 28/05/12. <http://hltshare.fbk.eu/EAMT2012/html/Papers/57.pdf>

A Phrase Table without Phrases: Rank Encoding for Better Phrase Table Compression. / Junczys-Dowmunt, Marcin.
16th Annual Conference of the European Association for Machine Translation (EAMT). ed. / Mauro Cettolo; Marcello Federico; Lucia Specia; Andy Way. Trento, Italy, 2012. p. 245-252.

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

TY - GEN

T1 - A Phrase Table without Phrases: Rank Encoding for Better Phrase Table Compression

AU - Junczys-Dowmunt, Marcin

PY - 2012

Y1 - 2012

N2 - This paper describes the first steps towards a minimum-size phrase table implementation to be used for phrase-based statistical machine translation. The focus lies on the size reduction of target language data in a phrase table. Rank Encoding (R-Enc), a novel method for the compression of word-aligned target language in phrase tables is presented. Combined with Huffman coding a relative size reduction of 56 percent for target phrase words and alignment data is achieved when compared to bare Huffman coding without R-Enc. In the context of the complete phrase table the size reduction is 22 percent.

AB - This paper describes the first steps towards a minimum-size phrase table implementation to be used for phrase-based statistical machine translation. The focus lies on the size reduction of target language data in a phrase table. Rank Encoding (R-Enc), a novel method for the compression of word-aligned target language in phrase tables is presented. Combined with Huffman coding a relative size reduction of 56 percent for target phrase words and alignment data is achieved when compared to bare Huffman coding without R-Enc. In the context of the complete phrase table the size reduction is 22 percent.

M3 - Conference contribution

SP - 245

EP - 252

BT - 16th Annual Conference of the European Association for Machine Translation (EAMT)

A2 - Cettolo, Mauro

A2 - Federico, Marcello

A2 - Specia, Lucia

A2 - Way, Andy

CY - Trento, Italy

T2 - 16th Annual Conference of the European Association for Machine Translation (EAMT)

Y2 - 28 May 2012 through 30 May 2012

ER -

A Phrase Table without Phrases: Rank Encoding for Better Phrase Table Compression

Abstract

Conference

Access to Document

Fingerprint

Cite this