Distributed Information Search and Retrieval for Astronomical Resource Discovery and Data Mining

Information search and retrieval has become by nature a distributed task. We look at tools and techniques which are of importance in this area. Current technological evolution can be summarized as the growing stability and cohesiveness of distributed architectures of searchable objects. The objects themselves are more often than not multimedia, including published articles or grey literature reports, yellow page services, image data, catalogs, presentation and online display materials, and “operations” information such as scheduling and publicly accessible proposal information. The evolution towards distributed architectures, protocols and formats, and the direction of our own work, are focussed on in this paper. 1. Needs and Existing Solutions in Astronomy Astronomy is characterized by massive data collections which are often incompatible, from many different sources, and stored in many different ways [1]. The user – astronomer or librarian – needs access in a common environment. The traditional solution to this issue is to have centralized databases available. This provides the common environment needed, and helps with quality control. Major databases can be considered, such as Simbad [2], NED [3] and LEDA [4] for astronomical objects, VizieR [5] for tables or ADS [6] for astronomical articles. Collecting data, storing them with one format in one database, and providing user interfaces, all require very considerable human resources. For the user, problems may arise when accessing data from different databases, because the data on a given subject reside on different servers with different metainformation and formatting. These problems are partially overcome by linkages between the data collections. For example, links between Simbad, NED, VizieR and ADS allow retrieval of articles based on an object which was cited. It is also possible to use a more direct protocol with NED and Simbad to retrieve information about an object, where one system can send a request to another and then format the response in an appropriate way. Such peerto-peer linkage is used by different servers, for example, to find all names for