Quality of indexing in library and information science databases

The study compares the quality of indexing in library and information science databases (Library Literature (LL), Library and Information Science Abstracts (LISA), and Information Science Abstracts (ISA)). An alternative method to traditional retrieval effectiveness tests, suggested by White and Griffith in their paper 'Quality of indexing in online databases' [13], is adopted to measure the quality of the controlled vocabulary of each database. The method involves identifying clusters of documents that are similar in content, searching for each document from a given cluster in a database, identifying the terms used by the databases to index each document, and calculating certain measures to determine the quality of indexing. Problems found with the White and Griffith discrimination index led the authors to propose an alternative discrimination index which takes into consideration the collection size of a database. Our analysis shows that LISA has the best quality of indexing out of the three databases.