Difference between revisions of "Leveraging Wikipedia and Context Features for Clinical Event Extraction from Mixed-Language Discharge Summary"
(+ wikilinks) |
(Infobox work) |
||
| Line 1: | Line 1: | ||
| + | {{Infobox work | ||
| + | | title = Leveraging Wikipedia and Context Features for Clinical Event Extraction from Mixed-Language Discharge Summary | ||
| + | | date = 2014 | ||
| + | | authors = [[Kwang-Yong Jeong]]<br />[[Wangjin Yi]]<br />[[Jae-Wook Seol]]<br />[[Jinwook Choi]]<br />[[Kyung-Soon Lee]] | ||
| + | | doi = 10.1007/978-3-319-12844-3_26 | ||
| + | | link = https://link.springer.com/chapter/10.1007%2F978-3-319-12844-3_26 | ||
| + | }} | ||
'''Leveraging Wikipedia and Context Features for Clinical Event Extraction from Mixed-Language Discharge Summary''' - scientific work related to [[Wikipedia quality]] published in 2014, written by [[Kwang-Yong Jeong]], [[Wangjin Yi]], [[Jae-Wook Seol]], [[Jinwook Choi]] and [[Kyung-Soon Lee]]. | '''Leveraging Wikipedia and Context Features for Clinical Event Extraction from Mixed-Language Discharge Summary''' - scientific work related to [[Wikipedia quality]] published in 2014, written by [[Kwang-Yong Jeong]], [[Wangjin Yi]], [[Jae-Wook Seol]], [[Jinwook Choi]] and [[Kyung-Soon Lee]]. | ||
== Overview == | == Overview == | ||
Unstructured clinical texts contain patients’ disease related narratives, but it is required elaborate work to mine the kind of information. Especially for the classification of semantic types of a clinical term, implementations of domain knowledge from resources such as the Unified Medical Language System (UMLS) are essential. The UMLS has a limitation in dealing with other languages. In this paper, authors leverage [[Wikipedia]] as well as UMLS for clinical event extraction, especially from clinical narratives written in mixed-language. Semantic [[features]] for clinical terms are extracted based on semantic networks of hierarchical [[categories]] in Wikipedia. Semantic types for Korean clinical terms are detected by using translation links and semantic networks in Wikipedia. An additional remarkable feature is a controlled vocabulary of clue words which can be contextual evidence to determine clinical semantic types of a word. The experimental result on 150 discharge summaries written in English and Korean showed 75.9% in F1-measure. This result shows that the proposed features are effective for clinical event extraction. | Unstructured clinical texts contain patients’ disease related narratives, but it is required elaborate work to mine the kind of information. Especially for the classification of semantic types of a clinical term, implementations of domain knowledge from resources such as the Unified Medical Language System (UMLS) are essential. The UMLS has a limitation in dealing with other languages. In this paper, authors leverage [[Wikipedia]] as well as UMLS for clinical event extraction, especially from clinical narratives written in mixed-language. Semantic [[features]] for clinical terms are extracted based on semantic networks of hierarchical [[categories]] in Wikipedia. Semantic types for Korean clinical terms are detected by using translation links and semantic networks in Wikipedia. An additional remarkable feature is a controlled vocabulary of clue words which can be contextual evidence to determine clinical semantic types of a word. The experimental result on 150 discharge summaries written in English and Korean showed 75.9% in F1-measure. This result shows that the proposed features are effective for clinical event extraction. | ||
Revision as of 14:16, 23 November 2019
Authors | Kwang-Yong Jeong Wangjin Yi Jae-Wook Seol Jinwook Choi Kyung-Soon Lee |
|---|---|
Publication date | 2014 |
DOI | 10.1007/978-3-319-12844-3_26 |
Links | Original |
Leveraging Wikipedia and Context Features for Clinical Event Extraction from Mixed-Language Discharge Summary - scientific work related to Wikipedia quality published in 2014, written by Kwang-Yong Jeong, Wangjin Yi, Jae-Wook Seol, Jinwook Choi and Kyung-Soon Lee.
Overview
Unstructured clinical texts contain patients’ disease related narratives, but it is required elaborate work to mine the kind of information. Especially for the classification of semantic types of a clinical term, implementations of domain knowledge from resources such as the Unified Medical Language System (UMLS) are essential. The UMLS has a limitation in dealing with other languages. In this paper, authors leverage Wikipedia as well as UMLS for clinical event extraction, especially from clinical narratives written in mixed-language. Semantic features for clinical terms are extracted based on semantic networks of hierarchical categories in Wikipedia. Semantic types for Korean clinical terms are detected by using translation links and semantic networks in Wikipedia. An additional remarkable feature is a controlled vocabulary of clue words which can be contextual evidence to determine clinical semantic types of a word. The experimental result on 150 discharge summaries written in English and Korean showed 75.9% in F1-measure. This result shows that the proposed features are effective for clinical event extraction.
