The viability of automatic indexing for biomedical literature

Automatic indexing is evaluated as an aid/replacement to manual indexing for biomedical literature. Manual indexing is costly and labour intensive. Technological innovations have the potential to increase efficiency and reduce costs. British Library produces a bibliographic database of allied and complementary medicine (AMED). This study compares articles which have been indexed manually for AMED with the same documents submitted to an automated indexing tool. The indexing tool selected was Helping Interdisciplinary Vocabulary Engineering, (HIVE) which is a jointly funded project by the University of North Carolina and the National Evolutionary Synthesis Center, North Carolina. A random selection of 100 records from a total of 1059 articles was selected. Each manually indexed document was compared with results returned by HIVE. Data analysis was made using SPSS. Results showed that HIVE does not provide a suitable replacement for the skills of a human indexer. Continued development of automatic indexing tools is recommended.