TY - GEN
T1 - Relation classification using coarse and fine-grained networks with SDP supervised key words selection
AU - Sun, Yiping
AU - Cui, Yu
AU - Hu, Jinglu
AU - Jia, Weijia
N1 - Funding Information:
Joint-project No. (025/2015/AMJ) of SAR Macau; University of Macau Funds Nos: CPG2018-00032-FST & SRG2018-00111-FST; Chinese National Research Fund (NSFC) Key Project No. 61532013; National China 973 Project No. 2015CB352401 and 985 Project of Shanghai Jiao Tong University: WF220103001. We also thank Xinsong ZHANG, Lester James V. Miranda and Mingyang YU for revising this paper.
Funding Information:
This work is supported by FDCT 0007/2018/A1, DCT-MoST Joint-project No. (025/2015/AMJ) of SAR Macau; University of Macau Funds Nos: CPG2018-00032-FST & SRG2018-00111-FST; Chinese National Research Fund (NSFC) Key Project No. 61532013; National China 973 Project No. 2015CB352401 and 985 Project of Shanghai Jiao Tong University: WF220103001. We also thank Xinsong ZHANG, Lester James V. Miranda and Mingyang YU for revising this paper.
Funding Information:
Acknowledgements. This work is supported by FDCT 0007/2018/A1, DCT-MoST
Publisher Copyright:
© 2018, Springer Nature Switzerland AG.
PY - 2018
Y1 - 2018
N2 - In relation classification, previous work focused on either whole sentence or key words, meeting problems when sentence contains noise or key words are extracted falsely. In this paper, we propose coarse and fine-grained networks for relation classification, which combine sentence and key words together to be more robust. Then, we propose a word selection network under shortest dependency path (SDP) supervision to select key words automatically instead of pre-processed key words and attention, which guides word selection network to a better feature space. A novel opposite loss is also proposed by pushing useful information in unselected words back to selected ones. In SemEval-2010 Task 8, results show that under the same features, proposed method outperforms state-of-the-art methods for relation classification.
AB - In relation classification, previous work focused on either whole sentence or key words, meeting problems when sentence contains noise or key words are extracted falsely. In this paper, we propose coarse and fine-grained networks for relation classification, which combine sentence and key words together to be more robust. Then, we propose a word selection network under shortest dependency path (SDP) supervision to select key words automatically instead of pre-processed key words and attention, which guides word selection network to a better feature space. A novel opposite loss is also proposed by pushing useful information in unselected words back to selected ones. In SemEval-2010 Task 8, results show that under the same features, proposed method outperforms state-of-the-art methods for relation classification.
KW - Coarse and fine-grained networks
KW - Opposite loss
KW - Relation classification
KW - Shortest dependency path
KW - selection
UR - http://www.scopus.com/inward/record.url?scp=85052229682&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85052229682&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-99365-2_46
DO - 10.1007/978-3-319-99365-2_46
M3 - Conference contribution
AN - SCOPUS:85052229682
SN - 9783319993645
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 514
EP - 522
BT - Knowledge Science, Engineering and Management - 11th International Conference, KSEM 2018, Proceedings
A2 - Liu, Weiru
A2 - Yang, Bo
A2 - Giunchiglia, Fausto
PB - Springer Verlag
T2 - 11th International Conference on Knowledge Science, Engineering and Management, KSEM 2018
Y2 - 17 August 2018 through 19 August 2018
ER -