�ݺ�ߣshows by User: JeeHyubKim

�ݺ�ߣshows by User: JeeHyubKim / http://www.slideshare.net/images/logo.gif �ݺ�ߣshows by User: JeeHyubKim / Thu, 06 Apr 2017 11:21:59 GMT �ݺ�ߣShare feed for �ݺ�ߣshows by User: JeeHyubKim Three hybrid classifiers for the detection of emotions in suicide notes /slideshow/three-hybrid-classifiers-for-the-detection-of-emotions-in-suicide-notes/74538004 i2b2challenge-170406112159
Suicides increasingly present a major concern in today's society. We describe our system for the recognition of emotions in suicide notes. Motivated by the sparse and imbalanced data as well as the complex annotation scheme, we have considered three hybrid approaches for distinguishing between the different categories. Each of the three approaches combines machine learning with manually derived rules, where the latter target very sparse emotion categories. The first approach considers the task as single label multi-class classification, where an SVM and a CRF classifier are trained to recognise fifteen different categories and their results are combined. Our second approach trains individual binary classifiers (SVM and CRF) for each of the fifteen sentence categories and returns the union of the classifiers as the final result. Finally, our third approach is a combination of binary and multi-class classifiers (SVM and CRF) trained on different subsets of the training data. We considered a number of different feature configurations. All three systems were tested on 300 unseen messages. Our second system had the best performance of the three,yielding an F1 score of 45.6% and a Precision of 60.1% whereas the best Recall (43.6%) was obtained using the third system.]]>
Suicides increasingly present a major concern in today's society. We describe our system for the recognition of emotions in suicide notes. Motivated by the sparse and imbalanced data as well as the complex annotation scheme, we have considered three hybrid approaches for distinguishing between the different categories. Each of the three approaches combines machine learning with manually derived rules, where the latter target very sparse emotion categories. The first approach considers the task as single label multi-class classification, where an SVM and a CRF classifier are trained to recognise fifteen different categories and their results are combined. Our second approach trains individual binary classifiers (SVM and CRF) for each of the fifteen sentence categories and returns the union of the classifiers as the final result. Finally, our third approach is a combination of binary and multi-class classifiers (SVM and CRF) trained on different subsets of the training data. We considered a number of different feature configurations. All three systems were tested on 300 unseen messages. Our second system had the best performance of the three,yielding an F1 score of 45.6% and a Precision of 60.1% whereas the best Recall (43.6%) was obtained using the third system.]]> Thu, 06 Apr 2017 11:21:59 GMT /slideshow/three-hybrid-classifiers-for-the-detection-of-emotions-in-suicide-notes/74538004 JeeHyubKim@slideshare.net(JeeHyubKim) Three hybrid classifiers for the detection of emotions in suicide notes JeeHyubKim Suicides increasingly present a major concern in today's society. We describe our system for the recognition of emotions in suicide notes. Motivated by the sparse and imbalanced data as well as the complex annotation scheme, we have considered three hybrid approaches for distinguishing between the different categories. Each of the three approaches combines machine learning with manually derived rules, where the latter target very sparse emotion categories. The first approach considers the task as single label multi-class classification, where an SVM and a CRF classifier are trained to recognise fifteen different categories and their results are combined. Our second approach trains individual binary classifiers (SVM and CRF) for each of the fifteen sentence categories and returns the union of the classifiers as the final result. Finally, our third approach is a combination of binary and multi-class classifiers (SVM and CRF) trained on different subsets of the training data. We considered a number of different feature configurations. All three systems were tested on 300 unseen messages. Our second system had the best performance of the three,yielding an F1 score of 45.6% and a Precision of 60.1% whereas the best Recall (43.6%) was obtained using the third system. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/i2b2challenge-170406112159-thumbnail.jpg?width=120&height=120&fit=bounds" /> Suicides increasingly present a major concern in today's society. We describe our system for the recognition of emotions in suicide notes. Motivated by the sparse and imbalanced data as well as the complex annotation scheme, we have considered three hybrid approaches for distinguishing between the different categories. Each of the three approaches combines machine learning with manually derived rules, where the latter target very sparse emotion categories. The first approach considers the task as single label multi-class classification, where an SVM and a CRF classifier are trained to recognise fifteen different categories and their results are combined. Our second approach trains individual binary classifiers (SVM and CRF) for each of the fifteen sentence categories and returns the union of the classifiers as the final result. Finally, our third approach is a combination of binary and multi-class classifiers (SVM and CRF) trained on different subsets of the training data. We considered a number of different feature configurations. All three systems were tested on 300 unseen messages. Our second system had the best performance of the three,yielding an F1 score of 45.6% and a Precision of 60.1% whereas the best Recall (43.6%) was obtained using the third system.

Three hybrid classifiers for the detection of emotions in suicide notes from Jee-Hyub Kim

]]> 149 2 https://cdn.slidesharecdn.com/ss_thumbnails/i2b2challenge-170406112159-thumbnail.jpg?width=120&height=120&fit=bounds document 000000 http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Learning to Extract Relations for Protein Annotation /JeeHyubKim/learning-to-extract-relations-for-protein-annotation presentation6-170309204133
Explains how to learn information extraction rules from unannotated corpora using inductive logic programming.]]>
Explains how to learn information extraction rules from unannotated corpora using inductive logic programming.]]> Thu, 09 Mar 2017 20:41:33 GMT /JeeHyubKim/learning-to-extract-relations-for-protein-annotation JeeHyubKim@slideshare.net(JeeHyubKim) Learning to Extract Relations for Protein Annotation JeeHyubKim Explains how to learn information extraction rules from unannotated corpora using inductive logic programming. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/presentation6-170309204133-thumbnail.jpg?width=120&height=120&fit=bounds" /> Explains how to learn information extraction rules from unannotated corpora using inductive logic programming.

Learning to Extract Relations for Protein Annotation from Jee-Hyub Kim

]]> 232 5 https://cdn.slidesharecdn.com/ss_thumbnails/presentation6-170309204133-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Literature Services Resource Description Framework /slideshow/literature-services-resource-description-framework/58650907 europepmcdataplatform-160224101220
Presented as part of EBI industrial workshop]]>
Presented as part of EBI industrial workshop]]> Wed, 24 Feb 2016 10:12:20 GMT /slideshow/literature-services-resource-description-framework/58650907 JeeHyubKim@slideshare.net(JeeHyubKim) Literature Services Resource Description Framework JeeHyubKim Presented as part of EBI industrial workshop <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/europepmcdataplatform-160224101220-thumbnail.jpg?width=120&height=120&fit=bounds" /> Presented as part of EBI industrial workshop

Literature Services Resource Description Framework from Jee-Hyub Kim

]]> 304 4 https://cdn.slidesharecdn.com/ss_thumbnails/europepmcdataplatform-160224101220-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Scalable Text Mining /slideshow/scalable-text-mining/58650622 scalabletextminingv32-160224100559
Good dictionaries are a key for text mining. We present an idea to build a platform where users can create their own dictionary and text-mining pipeline.]]>
Good dictionaries are a key for text mining. We present an idea to build a platform where users can create their own dictionary and text-mining pipeline.]]> Wed, 24 Feb 2016 10:05:59 GMT /slideshow/scalable-text-mining/58650622 JeeHyubKim@slideshare.net(JeeHyubKim) Scalable Text Mining JeeHyubKim Good dictionaries are a key for text mining. We present an idea to build a platform where users can create their own dictionary and text-mining pipeline. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/scalabletextminingv32-160224100559-thumbnail.jpg?width=120&height=120&fit=bounds" /> Good dictionaries are a key for text mining. We present an idea to build a platform where users can create their own dictionary and text-mining pipeline.

Scalable Text Mining from Jee-Hyub Kim

]]> 303 5 https://cdn.slidesharecdn.com/ss_thumbnails/scalabletextminingv32-160224100559-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Europe PubMed Central and Linked Data /slideshow/europe-pubmed-central-and-linked-data/53732574 epmcandlinkeddata-151009103046-lva1-app6891
On Europe PubMed Central, we extract identifies (e.g., accession numbers, data DOIs) in scientific articles. Recently, we started publishing mined identifiers on Linked Data Platform to improve the connectivity of our mined data.]]>
On Europe PubMed Central, we extract identifies (e.g., accession numbers, data DOIs) in scientific articles. Recently, we started publishing mined identifiers on Linked Data Platform to improve the connectivity of our mined data.]]> Fri, 09 Oct 2015 10:30:46 GMT /slideshow/europe-pubmed-central-and-linked-data/53732574 JeeHyubKim@slideshare.net(JeeHyubKim) Europe PubMed Central and Linked Data JeeHyubKim On Europe PubMed Central, we extract identifies (e.g., accession numbers, data DOIs) in scientific articles. Recently, we started publishing mined identifiers on Linked Data Platform to improve the connectivity of our mined data. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/epmcandlinkeddata-151009103046-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds" /> On Europe PubMed Central, we extract identifies (e.g., accession numbers, data DOIs) in scientific articles. Recently, we started publishing mined identifiers on Linked Data Platform to improve the connectivity of our mined data.

Europe PubMed Central and Linked Data from Jee-Hyub Kim

]]> 745 7 https://cdn.slidesharecdn.com/ss_thumbnails/epmcandlinkeddata-151009103046-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 https://cdn.slidesharecdn.com/profile-photo-JeeHyubKim-48x48.jpg?cb=1709542982 I'm a leading text-mining engineer with various experience in NLP, information retreival, text classification, and information extraction. I have a strong background in learning IE rules from unlabelled corpora. Besides text-mining, I have a good knowledge of linking text-mined results to knowlege bases on RDF platforms. Currently, I'm interested in learning deep learning and applying it to NLP problems. orcid.org/0000-0002-0359-2887 https://cdn.slidesharecdn.com/ss_thumbnails/i2b2challenge-170406112159-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/three-hybrid-classifiers-for-the-detection-of-emotions-in-suicide-notes/74538004 Three hybrid classifie... https://cdn.slidesharecdn.com/ss_thumbnails/presentation6-170309204133-thumbnail.jpg?width=320&height=320&fit=bounds JeeHyubKim/learning-to-extract-relations-for-protein-annotation Learning to Extract Re... https://cdn.slidesharecdn.com/ss_thumbnails/europepmcdataplatform-160224101220-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/literature-services-resource-description-framework/58650907 Literature Services Re...