Title: A method for extracting knowledge from medical texts including numerical representation

Authors: Kumiko Kiyoi, El-Sayed Atlam, Masao Fuketa, Tomoko Yoshinari, Jun-ichi Aoe

Addresses: Department of Information Science and Intelligent Systems, University of Tokushima, Tokushima 770-8506, Japan. ' Department of Information Science and Intelligent Systems, University of Tokushima, Tokushima 770-8506, Japan. ' Department of Information Science and Intelligent Systems, University of Tokushima, Tokushima 770-8506, Japan. ' Department of Information Science and Intelligent Systems, University of Tokushima, Tokushima 770-8506, Japan. ' Department of Information Science and Intelligent Systems, University of Tokushima, Tokushima 770-8506, Japan

Abstract: In recent years, there has been a tremendous growth of online text information related to digital libraries, medical diagnostic systems, remote education, news sources and electronic commerce. Numeric information is very important to understand numbers in texts of medical opinions. This paper presents a method for determining not only numbers but also expressions of modification to expand the range of corresponding numbers. The meaning of sentences is often determined by the combination of number expressions and their object words. Therefore, the presented method categorises with each meaning of expressions of modification and range expressions. According to experimental results for 948 Computer Tomography (CT) findings, the precision and recall for the extraction of number expressions are 98.23% and 97.62%, respectively. Moreover, the accuracy for the extraction of object words is 90.22%.

Keywords: knowledge extraction; numerical representation; multiattribute rules; medical texts; information retrieval; meaning; expressions.

DOI: 10.1504/IJCAT.2008.021945

International Journal of Computer Applications in Technology, 2008 Vol.33 No.2/3, pp.226 - 236

Available online: 10 Dec 2008 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article