Year: 2007 |
Download Free Adobe Acrobat Reader |
LHNCBC-2007-060 |
Extractive Summarization in Clinical Trials Protocol Summaries: A Case Study |
Rosemblat G, Graham L, Tse T |
In: Proceedings, 3rd Indian International Conference on Artificial Intelligence (IICAI-2007), pp. 1824-1837 |
This paper presents a method for extracting purpose statements from clinical trial protocol summaries. Simple summarization technology based on regular expressions and natural language processing techniques were applied in a controlled environment with structured sectioning to address an expressed user need: providing access to information about the purpose of specific clinical trials, originally in English, to Spanish speakers. Following an analysis of manually annotated data, based on a cascade of criteria, the Purpose Extractor Algorithm was developed to select tightly-focused candidate excerpts out of lengthy descriptions, for translation into Spanish. The extracts reduce the translation task and provide purposive content in biomedical text. The results were validated in a focused user study. It is anticipated that this extractive summarization approach may be generalized to documents from other databases as the algorithm can be tailored to different applications or needs. |
PDF |