The central project has the aim of allowing integrated searches on the databases of
GESIS member institutes under a common user friendly interface.
The impetus for the project GESINE (integrated social science information system) was an empirical survey of the information needs of users, in the form of an series of expert interviews at the GESIS institutes. The desire for a stronger integration of the differing information types was an overriding theme in these discussions. This would, for example, enable a researcher in electoral behaviour who was interested in obtaining information on party identification to obtain not only individual bibliographic and project references, but also survey data concerning attitudes of party members to their party as well as methodological information concerning the scales employed to measure identification with a particular party.
The identified information requirement led to the conception of an integrated information system for the social sciences based on the GESIS data. The technical-conceptual basis for this system is a relational data model based on distributed data sources. For the databases of the IZ (SOLIS and FORIS), the GESIS Clearinghouse and the ALLBUS Bibliography, such a data model had already been formulated and implemented in an ORACLE database.
The integration of structured and unstructured datasets has only recently become a concern of information science research. A general retrieval model for text and data does not exist - particularly not for the particular circumstances of the social sciences. The basis of such a model must therefore be borrowed from the domains and objectives of the particular application area.
As the technological basis for the underlying concept of an integrated search strategy for GESINE, quantitative-statistical retrieval models are suitable, which automatically index textual data and use ranking functionality in the output. In this regard, a comprehensive retrieval test of the intellectual indexing was compared at IZ with the automatic indexing carried out using the system freeWAIS-sf. It became evident that only multiple indexing leads to an effective increase in efficiency and performance.
Other areas which could benefit from automatic indexing are the GESIS Clearinghouse data sources, which arise in the implementation of the Layer Model (polycentric base model of information management).
In the discussions with GESIS members, it became apparent that an underlying problem in searching databases is the vagueness of the information. This is the case, for example, when the user, who has identified a particular project wishes to locate a project with a similar or related topic, or would like to obtain general overview literature on his chosen topic. In order to effectively support the user in such searches, components with added value are needed. These could identify the content and structural relationships between information objects and could therefore communicate the full complexity of the information. In this regard, the project was linked to the Project AKCESS (Assistance by Knowledge-Based Context Evaluation in Social Science Retrieval), in which solutions have already been found and successfully implemented for several spheres.
A further problem which arose during the internal GESIS information needs analysis was the support needed for repeated institutional work routines involving the entire GESIS membership. For this sphere, an example already exists in the component (COGET: computer-assisted production of documentations on social science topics), which can be employed to effectively produce topic based documentation.
Three sponsored projects or projects from the Hochschulsonderprogramm HSP (Higher Education Special Programmes) are relevant to these research and development activities:
© GESIS Peter Mutschke 1999-06-16