Prototype Geographic Search and Query Tools for a Library Digital Archive
By Jennifer N. Swift, Parisa Ghaemi, Daniel W. Goldberg, and John P. Wilson, University of Southern California

A GeoBrowser application, built by the University of Southern California (USC) GIS Research Laboratory for the USC Collection Information Systems Digital Information Management team, provides the existing archive Web site with advanced spatial search capabilities. The application uses El Segundo, California, as a demonstration area. Three new spatial search services were added to the existing Web site—a GeoBrowser (i.e., ArcIMS map), an address geocoder, and a gazetteer. These services act as filters that guide users in identifying and selecting library content in an area of interest. Currently, these spatial search tools can be used independently to search for library holdings.

GeoBrowser

The GeoBrowser application is hosted on a USC GIS Research Laboratory server.

The GeoBrowser is a custom-built Esri ArcIMS map coded using Java, JavaServer Pages, and JavaServer Faces (JSF). This application was developed and tested with Tomcat 4.1.29. The JSF API is used to render HTML pages, and the Esri ArcIMS Java Connector API is used to communicate with an Esri ArcIMS image service. The application also uses JavaScript and cascading style sheets (CSS) to provide the Web site with interactive GIS functionalities.

The GeoBrowser user interface provides standard navigation tool buttons (zoom in, zoom out, zoom global, fixed zoom in and out, and pan). The user draws a box on the map using the Search Box button to retrieve all the georeferenced records in the digital library database (Oracle Database Server) within the geographic boundaries of the box. The bounding coordinates of the search box (latitude and longitude of the northwest and southeast corners of the rectangle drawn on the map) can be viewed in text boxes below the map. A user may display the results of a search box search as points using the View USC Digital Archive Holdings tool. Selections often appear in clusters that may be better viewed by zooming in on the map. A user may also use the Clear Map View and Web Page tool to clear the search box and digital archive holdings from the map view and refresh the text-based search options. A Help tool is also available to view GeoBrowser tool button descriptions.

Map layers can be controlled by clicking the check boxes adjacent to each layer name. This displays or hides the layers in the map view. The list of layers shown beneath the map are point, line, and polygon shapefiles.

Geocoder

The geocoder feature was initially built to test the feasibility and scalability of new address matching procedures and algorithms developed by Rahul Bakshi and others at USC. The algorithm's main innovation is the ability to consult multiple sources to discover the actual number of addresses, their layout, and their actual geometry (i.e., features) occurring on a street, rather than using the address range associated with the segment for linear interpolation as traditional geocoders do. This additional knowledge allows for interpolation of locations along street segments with improved accuracy. These algorithms have been implemented into a scalable and reliable geocoding Web service, which is integrated into the GeoBrowser application.

The geocoder can be used to search the USC Digital Library for selections without requiring that the user know the spatial coordinates of an area of interest. A simple search based on address may be easier for a user who is unfamiliar with the geography of a region and cannot readily locate and draw a region of interest on a map.

Blue boxes are drawn using the Search Box button. Selected library records can be displayed using the View Library Holdings button. Search Box coordinates and visible map layers may be specified using the options below the map.

For the geocoder application to work efficiently, the user must enter both the street address and ZIP Code. An address and ZIP Code search can be defined by entering any address in the United States to search for selections within a given radius of the address. The radius has been predefined in the application by the Digital Library staff. The ZIP Code may narrow the geocoding search to a smaller area than a search on a street name. The search has the greatest chance of success if the user can also provide the ZIP Code in those instances in which there are many street segments with the same name. The search is executed by clicking the red Search button beneath the text-based Advanced Search options. All entries and selections in the text-based search tools can be cleared by clicking the red Reset button.

Gazetteer

The gazetteer database and service are restricted to the city of El Segundo in Los Angeles County. The gazetteer was built to demonstrate how newly developed Web harvesting tools could be deployed to build gazetteer databases quickly and cheaply. The gazetteer provides a highly detailed location filter that the user employs to identify a specific location, such as a business, and retrieve all of the Digital Library selections within a specified geographic buffer around the chosen location. The buffer radius is currently predefined in the application by the Digital Library staff.

Continued on page 2