| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a medical information extraction device and program capable of learning from information on the correlation between dominant and subordinate concepts where disease names are the dominant concepts while items related to the disease names are subordinate concepts, and on digitized documents, extracting new subordinate concepts, and updating a list of subordinate concepts. <P>SOLUTION: The CPU 10 of the medical information extraction device 1 extracts passages related to medical practice from Web documents and calculates the degree of correlation between each item in a list of the related items and the name of a disease to be processed, based on a target set of documents including the passages extracted. Then, based on the target set of documents including the passages extracted, the CPU 10 extracts new items related to the name of the disease to be processed and not included in the list, and adds the new items to the list. <P>COPYRIGHT: (C)2008,JPO&INPIT |
