Joohong Lee roomylee. Tensorflow Implementation of Convolutional Neural Network for Relation Extraction (COLING 2014, NAACL 2015) Python 202 57. Network Affairs Committee Byung Heon Lee (Kyungpook Nat’l Univ., Korea) Ethic Committee Jung Weon Lee (Seoul Nat’l Univ., Korea) Ilchun Committee Jong-Il Kim (Seoul Nat’l Univ., College of Medicine, Korea) Welfare Committee Dae-Jin Yun (Konkuk Nat’l Univ., Korea) Committee for Advancement Joohong Ahnn (Hanyang Univ., Korea) Winter.
An engaged lifestyle is seen as an important component of successful ageing. Many older adults with high participation in social and leisure activities report positive wellbeing, a fact that fuelled the original activity theory and that continues to influence researchers, theorists and practitioners.
Abstract: Classifying semantic relations between entity pairs in sentences is animportant task in Natural Language Processing (NLP). Most previous models forrelation classification rely on the high-level lexical and syntactic featuresobtained by NLP tools such as WordNet, dependency parser, part-of-speech (POS)tagger, and named entity recognizers (NER). In addition, state-of-the-artneural models based on attention mechanisms do not fully utilize information ofentity that may be the most crucial features for relation classification. Toaddress these issues, we propose a novel end-to-end recurrent neural modelwhich incorporates an entity-aware attention mechanism with a latent entitytyping (LET) method. Our model not only utilizes entities and their latenttypes as features effectively but also is more interpretable by visualizingattention mechanisms applied to our model and results of LET. Experimentalresults on the SemEval-2010 Task 8, one of the most popular relationclassification task, demonstrate that our model outperforms existingstate-of-the-art models without any high-level features.
Submission history
From: Joohong Lee [view email][v1]Wed, 23 Jan 2019 23:19:45 UTC (1,032 KB)
Full-text links:
Download:
Current browse context:
References & Citations
DBLP - CS Bibliography
Sangwoo Seo
Yong Suk Choi
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs and how to get involved.
Abstract: Classifying semantic relations between entity pairs in sentences is animportant task in Natural Language Processing (NLP). Most previous models forrelation classification rely on the high-level lexical and syntactic featuresobtained by NLP tools such as WordNet, dependency parser, part-of-speech (POS)tagger, and named entity recognizers (NER). In addition, state-of-the-artneural models based on attention mechanisms do not fully utilize information ofentity that may be the most crucial features for relation classification. Toaddress these issues, we propose a novel end-to-end recurrent neural modelwhich incorporates an entity-aware attention mechanism with a latent entitytyping (LET) method. Our model not only utilizes entities and their latenttypes as features effectively but also is more interpretable by visualizingattention mechanisms applied to our model and results of LET. Experimentalresults on the SemEval-2010 Task 8, one of the most popular relationclassification task, demonstrate that our model outperforms existingstate-of-the-art models without any high-level features.
Submission history
From: Joohong Lee [view email][v1]
Joohong Network Connection
Wed, 23 Jan 2019 23:19:45 UTC (1,032 KB)Full-text links:
Download:
Current browse context:References & Citations
DBLP - CS Bibliography
Sangwoo Seo
Yong Suk Choi
Joohong Network Services
Joohong Network Directv
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Joohong Network Provider
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs and how to get involved.
Joohong Network Login