Executive  summaryChapter1Chapter2Chapter3
Chapter4Chapter5Chapter6Chapter7Home

APPENDIX 3F—FUNCTIONALITY

American

Procedures: Both the dynamically created screens and the searching capabilities were implemented through CGI scripts written in Perl on the UNIX server. The user could browse through collections by period looking at thumbnail images, which then had links to the full data records. They could also do fielded searches or simple searches by all fields for a specified keyword. The available Boolean operators were "and", "or", and "not". The options for display of the results were a mid-size image of the artwork, the type of object, the credit line and label data fields. The user could click on an image to get larger sizes.

Columbia

Procedures: The index was created by Glimpse, a search engine package. Their system allowed for full-text searching of object descriptions only. The types of searching available were Boolean, fuzzy match and case sensitivity in all fields. The display of a search result was text with the museum name and object title/caption highlighted. The skills primarily needed were database development, and knowledge of C, Perl and shell scripts. In Year 2 a programmer/Web specialist was added to the staff.

Cornell

Procedures: They used the search engine built into FileMaker Pro. The database was accessed from the Web using ROFM CGI shareware. Both fielded searches and generic keyword searching were supported. Only the Boolean operator "and" was supported. Other features included searching by date range and exact match, and links to definitions of the MESL data dictionary terms. The display options for search results were either a thumbnail image with brief text or text only. The user could modify both the number of records and/or images displayed, and sort the result set.

Illinois

Procedures: They implemented a search engine that used a web-based form interface to query an SQL-server database through CGI-scripts written in C++. Fielded searches were not offered. The search interface was a single line for input. The entered terms were searched for in each of the indexed fields (when multiple words were entered, a default Boolean "and" was inserted). There were two options for displaying search results: non-frame and frame. The non-frame version displayed up to 10 thumbnails and brief indexed record; the user could click on the image and retrieve a medium-size version. The user was also able to retrieve all 32 MESL dictionary fields. The frame version allowed for the same display as the non-frame display, but when the user clicked to retrieve the medium-size image the image was displayed in the same browser screen, but in a different frame. The medium-size images and up to 10 thumbnails can be displayed simultaneously.

Maryland

Procedures: They provided browsing capabilities in both their web-based and their SearchSlide™ implementations. Their web-based implementation allowed for searches by artist, title and keyword and it was case sensitive. The SearchSlide™ implementation allowed for very complex Boolean searches. The simple query function was a Boolean "and" of all search fields selected. There were two search results display options, one was to view a thumbnail and its accompanying text, and the other was to display the results as a screen of thumbnails. An additional functionality that they were able to make available through SearchSlide™ was the ability to create an "electronic carousels" of images; that is, to create a left-screen/right-screen list of objects. These images could then be saved and projected in the classroom side by side via ProjectSlide™.

Michigan

Procedures: They used a custom-developed search engine, Full Text Lexicographer, which offered many complex search features such as: Boolean "and", "or", "not" and "and-not" searches, proximity searching, fielded searches, automatic stemming, and searches by numeric ranges. Full Text Lexicographer offered incremental updates and deletions, and it was usable from the command line or as a back end to other programs. The search results were sortable by any searchable field, and display options included thumbnail view or text list, and all structured data viewable for all records. In addition, they provided a table of searchable terms and phrases and a table of special character substitutions.

Virginia

Procedures: They used Open Text technology, a sophisticated high-speed searching tool, for indexing, searching and retrieval. Perl scripts were written to generate the web pages on the fly. Their search interface offered fielded searches. After completing a search, the user was presented with an intermediate screen which listed the number of results. The user could choose from three types of displays: thumbnail with brief record (linked to medium sized image and full record), thumbnail with checkbox (user checks images for subsequent display of the brief record), and brief record only (the user then selects records to see thumbnail view).

Back to Chapter 3

 


The Cost of Digital Image Distribution:
The Social and Economic Implications of
the Production, Distribution, and Usage of Image Data

By Howard Besser & Robert Yamashita
http://sunsite.berkeley.edu/Imaging/Databases/1998mellon