C. S. Grant, A. Accomazzi, G. Eichhorn, M. J. Kurtz, and S. S. Murray
Smithsonian Astrophysical Observatory, 60 Garden St., Cambridge, MA 02138
The ADS Mosaic Catalog Service currently provides access to about 150 data catalogs located at various nodes across the country. This service was developed to provide a simple means of making on-line catalogs available for searching. Sites with catalogs already in a relational database need only install a server which we provide and create a data dictionary with tools we will provide.
In the current implementation of the service, users may search on any field contained within the catalog, either by typing in a Standard Query Language (SQL) request, or by filling in a table template (QBT). In order to provide the capability of searching catalogs by position, our software needs to be able to identify what kinds of positions are contained in which columns. Because we do not maintain the data, we have no control over what coordinate types and units are in the catalogs. We therefore propose a standard method of identifying column contents with an associated data dictionary file. The data dictionary can also be extended beyond positional column identification and used to key on any astronomically-interesting field (such as magnitude or class).
We propose having one file per database which lists information about each table in the database ( Database Data Dictionary). In addition, we propose having one file per table which lists information about each column in the table ( Table Data Dictionary). Files will be in FITS-like format, with keyword=value pairs. We will provide software which creates these files from the catalog documentation file. Nodes will maintain the files so that updates to the databases can be handled without requiring ADS intervention.
Both the descriptions of the catalogs and the keywords assigned to the catalogs are WAIS-indexed. This allows users to search the catalog content for information about the catalogs (such as which catalogs contain redshift).
There is one Database Data Dictionary file per database at a node. This file is used by the software to construct lists of tables for selection (``by node" or ``by subject"). The field specifications are detailed in Table 1.
Table 1: Database Data Dictionary
The subjects are used for grouping similar catalogs together. The keywords are used for WAIS searches to enable users to find a catalog about a particular subject. Both of these are to be taken from word sets already in use by the ADS project. The software provided to help generate the data dictionary files will facilitate the selection of appropriate subjects and keywords.
There is one Table Data Dictionary per table at a node. The counter, nnn, specifies the column number in the table. Additional keywords (such as NAXIS1 and NAXIS2) are added when a FITS table is written out containing results of a query. The field specifications are detailed in Table 2.
Table 2: Table Data Dictionary
The keywords which begin with ``DB_" correspond to keywords in the associated Database Data Dictionary. These are essentially a header to the data dictionary, giving information about the table (as opposed to information about the columns). TTYPE and TFORM are the only two required keywords for each column. Unused keywords may be omitted.
There are two keywords in the Table Data Dictionary which indicate to the server software that the corresponding column contains special attributes. The data flag, TDFLGnnn, describes the type of attribute (such as position or image); the auxiliary flag, TAFLGnnn, describes additional information associated with the data flag (such as coordinate specifications or a directory pathname). Table 3 lists examples of data flags.
Positional data flags indicate to the server software that the corresponding column contains coordinates. The primary set of positions to be used in searching (determined by the data providers) is flagged with a data flag ``p". There can be only one set of primary positions, but there can be multiple secondary position data flags. The positional auxiliary flags (see Table 4 for examples) describe what kind of coordinates are represented in a colon-separated list, including: which coordinate (x or y), the coordinate units (DD, RAD, sex1, sex2, etc.), the coordinate name where appropriate (some sexagesimal representations), the coordinate system (EQ, GAL, or ECL), and the coordinate epoch where appropriate (B1950 or J2000).
Table 3: Data Flags
Table 4: Positional Auxiliary Flags
The remaining data flags give tremendous flexibility in the search capabilities. For example, by flagging which column contains the name of the associated image, data archives can be built to retrieve lists of images available in a given catalog. Likewise, by flagging which column is a magnitude column, plot routines can be automated to size the plot symbols based on magnitude. Other data flags can be added by the nodes as desired.
The ADS Mosaic Catalog Service offers data sites the opportunity to make their data available from a centralized location. We provide the software to search catalogs individually, and once the data dictionaries are in place, we will implement positional searches as well as multiple catalog searches. Output from the Catalog Service may be returned in a variety of formats (including ASCII, FITS, and ADS tables) giving users the necessary flexibility for manipulation of their results.
Putting a catalog on-line requires only four steps: (1) running an httpd, (2) installing SQL software (we will assist), (3) writing catalog documentation, and (4) creating data dictionary (we will assist). The ADS provides WWW access to catalogs while still allowing the nodes to maintain ownership of all associated files for updating and maintenance. We offer a homogeneous Mosaic-based query mechanism which is already in place and ready to be expanded. For more information, please contact adscfa.harvard.edu.
This project is funded by the NASA Astrophysics program under grant NCCW--0024.