Abstract
We present the initial results of our experiments on document alignment for the online news domain. Specifically, as apposed to cross-site comparable news alignment, we focus on the identification of parallel documents from within the same multilingual websites. In such a setting parallel news stories oftentimes turn out to be direct translations of each other with a tendency of sharing common media and displaying proximity in publication date. We leverage this domain-specific property of the data and propose a straightforward yet competitive heuristic that performs on par with a machine learning-based method in terms of precision, and outperforms a widely used bitext extraction system on a range of metrics. Moreover, this heuristic has allowed us to identify comparable documents overlooked by a human annotator. Although both rule-and learning-based methods that we present are language independent, we specifically focus on the Russian-Kazakh language pair as the present study is one of the initial steps towards a greater objective of building a corresponding parallel corpus and a machine translation system.
| Original language | English |
|---|---|
| Title of host publication | Application of Information and Communication Technologies, AICT 2016 - Conference Proceedings |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| ISBN (Electronic) | 9781509018406 |
| DOIs | |
| Publication status | Published - Jul 25 2017 |
| Event | 10th IEEE International Conference on Application of Information and Communication Technologies, AICT 2016 - Baku, Azerbaijan Duration: Oct 12 2016 → Oct 14 2016 |
Conference
| Conference | 10th IEEE International Conference on Application of Information and Communication Technologies, AICT 2016 |
|---|---|
| Country | Azerbaijan |
| City | Baku |
| Period | 10/12/16 → 10/14/16 |
Fingerprint
Keywords
- Document alignment
- machine translation
- parallel corpus
ASJC Scopus subject areas
- Computer Vision and Pattern Recognition
- Computer Science Applications
- Computer Networks and Communications
- Information Systems
- Modelling and Simulation
Cite this
Identification of the parallel documents from multilingual news websites. / Myrzakhmetov, Bagdat; Sultangazina, Aitolkyn; Makazhanov, Aibek.
Application of Information and Communication Technologies, AICT 2016 - Conference Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. 7991684.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Identification of the parallel documents from multilingual news websites
AU - Myrzakhmetov, Bagdat
AU - Sultangazina, Aitolkyn
AU - Makazhanov, Aibek
PY - 2017/7/25
Y1 - 2017/7/25
N2 - We present the initial results of our experiments on document alignment for the online news domain. Specifically, as apposed to cross-site comparable news alignment, we focus on the identification of parallel documents from within the same multilingual websites. In such a setting parallel news stories oftentimes turn out to be direct translations of each other with a tendency of sharing common media and displaying proximity in publication date. We leverage this domain-specific property of the data and propose a straightforward yet competitive heuristic that performs on par with a machine learning-based method in terms of precision, and outperforms a widely used bitext extraction system on a range of metrics. Moreover, this heuristic has allowed us to identify comparable documents overlooked by a human annotator. Although both rule-and learning-based methods that we present are language independent, we specifically focus on the Russian-Kazakh language pair as the present study is one of the initial steps towards a greater objective of building a corresponding parallel corpus and a machine translation system.
AB - We present the initial results of our experiments on document alignment for the online news domain. Specifically, as apposed to cross-site comparable news alignment, we focus on the identification of parallel documents from within the same multilingual websites. In such a setting parallel news stories oftentimes turn out to be direct translations of each other with a tendency of sharing common media and displaying proximity in publication date. We leverage this domain-specific property of the data and propose a straightforward yet competitive heuristic that performs on par with a machine learning-based method in terms of precision, and outperforms a widely used bitext extraction system on a range of metrics. Moreover, this heuristic has allowed us to identify comparable documents overlooked by a human annotator. Although both rule-and learning-based methods that we present are language independent, we specifically focus on the Russian-Kazakh language pair as the present study is one of the initial steps towards a greater objective of building a corresponding parallel corpus and a machine translation system.
KW - Document alignment
KW - machine translation
KW - parallel corpus
UR - http://www.scopus.com/inward/record.url?scp=85034235250&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85034235250&partnerID=8YFLogxK
U2 - 10.1109/ICAICT.2016.7991684
DO - 10.1109/ICAICT.2016.7991684
M3 - Conference contribution
BT - Application of Information and Communication Technologies, AICT 2016 - Conference Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
ER -