A translating computer interface for a network of heterogeneous interactive information retrieval systems

The need for a network of heterogeneous interactive bibliographic information retrieval systems is projected from the facts of wide acceptance and growing demand for these systems and the limitations of their use caused by limited online data base size. Because of the established characters of the different I-R systems and unlikelihood that a standardized I-R system leading directly to the ultimate uniform network will soon be generally adopted, we propose an intermediate solution in which computer interfaces would provide the networking capability by translating and converting among the diverse languages and data bases of existing systems.We have begun work on such a network in which the computer interface is based on the concept of a common language for commands, indexing vocabularies, and data base structures. In particular, the common language for commands and data base structures is based on identifying the basic or primitive I-R functions and bibliographic data base elements. The conversion among indexing vocabularies is based on the concept of a Master Index and Thesaurus containing the conglomerate thesaurus information from the separate data bases. In addition, the phrase decomposition and stemming of the individual words in the search request and subject index phrases are used as further techniques for automated conversion among diverse vocabularies. An initial experimental interface is described which interconnects the M.I.T. Intrex system, the MEDLINE retrieval system, and the TYMNET computer network using the ARPANET Terminal Interface Processors for intercomputer communication.