A signature file scheme based on multiple organizations for indexing very large text databases

• Published in: Journal of the American Society for Information Science Vol. 41; no. 7; pp. 508 - 534
• Main Authors: Kent, A., Sacks-Davis, R., Ramamohanarao, K.
• Format: Journal Article
• Language: English
• Published: Washington, D.C Wiley Subscription Services, Inc., A Wiley Company 01.10.1990; John Wiley & Sons; American Documentation Institute; Wiley Periodicals Inc
• Subjects: Access to information; Bibliographic Databases; Computerized information storage and retrieval; Computerized subject indexing; Data Files; Exact sciences and technology; File organization; Formats. Markup languages. Codification. Conversion; Full Text Databases; Indexing; Information and communication sciences; Information and document structure and analysis; Information processing and retrieval; Information Retrieval; Information science. Documentation; Information storage and retrieval; Information work; International conferences; Libraries; Library Research; Logic programming; Mathematical Models; Methods; Office automation; Relational databases; Sciences and techniques of general use; Signature Files; Signatures; Subject Index Terms; Subject indexing; New York; United States > US; Large dimension; Inverted file; Query; Information conversion; Data storage; Mathematical method; Document access; Addressing; Implementation; File layout; Research and development; Automated processing; Documentation data processing; Signing; Database; Hashing; Question answering; Full text; Indexing
• ISSN: 0002-8231; 1097-4571
• DOI: 10.1002/(SICI)1097-4571(199010)41:7<508::AID-ASI5>3.0.CO;2-J

A new signature file method for accessing information from large databases containing both formatted and free text data is presented. The new method, called the multiorganizational scheme is proposed for indexing very large databases containing hundreds of thousands or possibly millions of records. With this method, records are grouped into blocks and signatures are formed for each block of records. These signatures are stored in a block descriptor file using a storage device called the bit slice organization. By forming multiple block descriptor files, each based on a possibly different grouping of records into blocks, it is possible to efficiently determine record matches on query. Both computational results based on a mathematical model as well as experimental results using a library database are presented. These results show that the method provides effective access to large text databases. © 1990 John Wiley & Sons, Inc.