ABSTRACT
The strategies for automatic extraction of key words from medical news were put forward by adding the MeSH terms into the general classification table in combination with the length of MeSH terms and location-weigh-ted MeSH terms.The key words randomly selected from 100 papers reporting medical news on 10 Websites were in-dexed and verified by machine indexing.The extraction accuracy was 0.34 and the recall rate was 0.30, showing that the strategies can be used for automatic extraction of key words from medical news.
ABSTRACT
After the necessity of using medical news information and the advances in its automatic indexing were analyzed, a novel automatic controlled indexing method of medical news text was put forward. The method intro-duced translated MeSH vocabulary as the main indexing words, merging Chinese commonly used word segmentation dictionary, then calculated word frequency for document text which added split token and sorted it, choose top 5 high-frequency words in MeSH vocabulary indexed document after deleting high-frequency words not in MeSH vo-cabulary.
ABSTRACT
OBJECTIVE: Indexing medical documents is important to retrieve medical information efficiently, but it is labor intensive and an annoying task for indexers or authors. This paper presents that whether an automatic indexing program can help the human task for Korean medical keyword indexing. METHODS: We developed an automatic indexing program using Korean Medical Subject Heading(K-MeSH) and evaluated the performance as compared with technical indexers and authors. RESULTS: Experimental result was that the current program's performance was much lower than technical indexers', but it was same as the authors' performance. CONCLUSION: The result showed that it is very affirmative to develop the automatic indexing program to help authors at least, and to help technical indexers with improving the program by enriching K-MeSH and utilizing K-MeSH structure.