On the architecture of a system integrating data base management and information retrieval

The data model, i.e. data structures and operations needed for a system integrating the management of formated textual data (DBMIRS) are discussed. It is investigated how this data model fits into the ANSI-SPARC three schema architecture for data base management systems. The conclusion is that the DBMIRS should be regarded to be a new external data model. This would require only small changes to the concepts discussed so far for the conceptual and internal level. The advantages of this approach are:- The dichotomy between IRS and DBMS does not further exist.- The main part of the software can be used for IR systems as well as DB systems.- IR research can concentrate on the problems inherent in the retrieval of texts, e.g. automatic indexing, relevance feedback techniques.