Statistical Analysis on Indexing Depth in INIS

Abstract

The indexing tendency in INIS has been analyzed by a statistical means. In this analysis were counted the indexer assigned descriptors per piece of literature and the computer generated broader terms per indexer assigned descriptor, and also the number of split documents. The INIS employs a decentralized input system, and further, the indexing itself differs with the respective subject categories. The numbers of descriptors and broader terms and of split documents were thus obtained for each inputting country and subject category. The difference in indexing among the countries for given subject category was considerably large, far more than expected. The indexing depth is then related not only with the number of descriptors per piece of literature, it is also largely influenced by the ratio of computer- and indexer-assigned descriptors.