Can syntax help? Improving an LSTM-based Sentence Compression Model for New Domains

• المؤلفون الرئيسيون: WANG, Liangguo, JIANG, Jing, CHIEU, Hai Leong, ONG, Chen Hui, SONG, Dandan, LIAO, Lejian
• التنسيق: text
• اللغة: English
• منشور في: Institutional Knowledge at Singapore Management University 2017
• الموضوعات: Databases and Information Systems; Programming Languages and Compilers
• الوصول للمادة أونلاين: https://ink.library.smu.edu.sg/sis_research/3901; https://ink.library.smu.edu.sg/context/sis_research/article/4903/viewcontent/P17_1127.pdf

In this paper, we study how to improve thedomain adaptability of a deletion-basedLong Short-Term Memory (LSTM) neuralnetwork model for sentence compression.We hypothesize that syntactic informationhelps in making such modelsmore robust across domains. We proposetwo major changes to the model: usingexplicit syntactic features and introducingsyntactic constraints through Integer LinearProgramming (ILP). Our evaluationshows that the proposed model works betterthan the original model as well as a traditionalnon-neural-network-based modelin a cross-domain setting.