Extracting Specific Medical Data Using Semantic Structures

Denecke, Kerstin; Bernauer, Jochen

doi:10.1007/978-3-540-73599-1_35

Cited by 1 publication

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Even if this information is not available, most users will point it out in the review text. In fact, many hidden clues can be found by using information extraction techniques (Denecke & Bernauer, 2007).…”

Section: Extraction Of Contextual Cluesmentioning

confidence: 99%

A Semantic Framework based on Domain Knowledge for Opinion Mining of Drug Reviews

Noferesti

Shamsfard

2022

JART

View full text Add to dashboard Cite

Opinion mining has been attracting increasing attention in recent years. Existing approaches to opinion mining that have worked on general domains face two major challenges for polarity classification of drug reviews. Firstly, indirect opinions frequently occur in the drug domain, while the existing methods have mainly focused on direct opinions and ignored the indirect ones. Secondly, previous works are not sufficient for polarity classification of ambiguous concepts in the drug domain. This paper proposed a semantic framework based on domain knowledge for resource construction and exploitation for indirect opinion mining of drug reviews. Accordingly, some methods were introduced, developed, and compared for building and exploiting a combined knowledge base, polarity-tagged corpus, and context-aware resources for the polarity detection of drug reviews. The test results showed that the proposed methods reached a precision of 89.18% and 80.4% in the application of the combined knowledge base and the polarity-tagged corpus for polarity detection of indirect opinions, respectively. Also, a precision of 79.93% was achieved with the use of context-aware resources constructed for the polarity detection of ambiguous concepts. Overall, the results obtained demonstrated the performance of the proposed methods compared to the existing methods.

show abstract

Section: Extraction Of Contextual Cluesmentioning

confidence: 99%