Abstract
The effective use of digital libraries demands maintenance of bibliographic databases. Useful bibliographic information appears in the reference fields of academic papers, so we are developing a method for automatic extraction of bibliographic information from reference strings using a conditional random field (CRF). However, at least a few hundred reference strings are necessary to learn an accurate CRF. In this paper, we propose active learning and transfer learning techniques to reduce the required training data for CRFs. We evaluate extraction accuracies and the associated training cost by experiments.
| Original language | English |
|---|---|
| Title of host publication | 2017 12th International Conference on Digital Information Management, ICDIM 2017 |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| Pages | 59-64 |
| Number of pages | 6 |
| Volume | 2018-January |
| ISBN (Electronic) | 9781538606643 |
| DOIs | |
| Publication status | Published - Jan 2 2018 |
| Event | 12th International Conference on Digital Information Management, ICDIM 2017 - Fukuoka, Japan Duration: Sep 12 2017 → Sep 14 2017 |
Other
| Other | 12th International Conference on Digital Information Management, ICDIM 2017 |
|---|---|
| Country | Japan |
| City | Fukuoka |
| Period | 9/12/17 → 9/14/17 |
Fingerprint
Keywords
- active learning
- bibliography extraction
- confidence measure
- CRF
- transfer learning
ASJC Scopus subject areas
- Computer Networks and Communications
- Information Systems
- Information Systems and Management
Cite this
CRF-based bibliography extraction from reference strings using a small amount of training data. / Namikoshi, Daiki; Ohta, Manabu; Takasu, Atsuhiro; Adachi, Jun.
2017 12th International Conference on Digital Information Management, ICDIM 2017. Vol. 2018-January Institute of Electrical and Electronics Engineers Inc., 2018. p. 59-64.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - CRF-based bibliography extraction from reference strings using a small amount of training data
AU - Namikoshi, Daiki
AU - Ohta, Manabu
AU - Takasu, Atsuhiro
AU - Adachi, Jun
PY - 2018/1/2
Y1 - 2018/1/2
N2 - The effective use of digital libraries demands maintenance of bibliographic databases. Useful bibliographic information appears in the reference fields of academic papers, so we are developing a method for automatic extraction of bibliographic information from reference strings using a conditional random field (CRF). However, at least a few hundred reference strings are necessary to learn an accurate CRF. In this paper, we propose active learning and transfer learning techniques to reduce the required training data for CRFs. We evaluate extraction accuracies and the associated training cost by experiments.
AB - The effective use of digital libraries demands maintenance of bibliographic databases. Useful bibliographic information appears in the reference fields of academic papers, so we are developing a method for automatic extraction of bibliographic information from reference strings using a conditional random field (CRF). However, at least a few hundred reference strings are necessary to learn an accurate CRF. In this paper, we propose active learning and transfer learning techniques to reduce the required training data for CRFs. We evaluate extraction accuracies and the associated training cost by experiments.
KW - active learning
KW - bibliography extraction
KW - confidence measure
KW - CRF
KW - transfer learning
UR - http://www.scopus.com/inward/record.url?scp=85049371656&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85049371656&partnerID=8YFLogxK
U2 - 10.1109/ICDIM.2017.8244665
DO - 10.1109/ICDIM.2017.8244665
M3 - Conference contribution
AN - SCOPUS:85049371656
VL - 2018-January
SP - 59
EP - 64
BT - 2017 12th International Conference on Digital Information Management, ICDIM 2017
PB - Institute of Electrical and Electronics Engineers Inc.
ER -