Past meeting: October 4, 1997
Classification Techniques
Twenty-four ASI/PNW members and non-members compared the classification methods of three local organizations at the chapter's second meeting of the year in Olympia, Washington on Saturday, October 4.
Keyworder Lee Lawing of PhotoDisc, a Seattle-based company which has published more than 50,000 photographs on CD-ROM and the Web since 1991, demonstrated the three-step process he uses to help customers find a particular image. The goal is to provide as many options as possible for customers to find the same photo.
- Determine the main category in which a photograph appears when customers browse PhotoDisc's collections.
- Assign subcategories that are appropriate to the same photograph. Customers can also browse these subcategories.
- Add keywords which users can type into a dialog box to search for a photo. Lawing uses three types of keywords: general, basic keywords used to describe the topic of the image; free-form, keywords specific to a photo; and concepts, keywords capturing a feeling or idea portrayed by the image.
Lawing's tasks seem straightforward when compared to the classification methods at the Washington State Archives in Olympia. State mandates require the 50,000 cubic feet of records housed there to be organized in a variety of ways, including the names of the government agencies and the subject terms they use to classify their documents, says Archivist David Hastings. Records are also indexed by the name of a series (such as a study on the effects of downwind radiation at Hanford Nuclear Reservation), the volume number, date or time period, and geographic location.
Because classification methods vary for each document, narratives are written for all records to help the archives' four staff members find information more quickly. In addition, when researchers visit the archives, they are interviewed so that staff can direct them to the records appropriate to their topics. The most frequent visitors are lawyers looking for "legislative intent" (the reason a law was passed), genealogists, and historians, Hastings said.
The archives uses a GenCat database to store and search its records.
Full-text searching is the primary keywording technique used at Professional Library Services in Seattle, which gathers, retrieves, and organizes information for corporate libraries. Documents are scanned into text files and given to customers on computer disks or CD-ROMs, along with the software that allows them to search for information. The advantage of electronic information, according to David Pearlstein, Director of Research and Marketing, is that it can be searched quickly, but full-text indexing is not always accurate. Fuzzy searches result in hits that don't match the keyword entered, because the software may not recognize certain characters clearly. For example, the lower-case letter "f" may be mistaken for "t," or the word "government" may look like "govemment" because the letters "r," "n," and "m" may be kerned too tightly. Also, handwritten documents can't be searched, because the software can't recognize the words.
Skills needed to work at these three companies are as diverse as the classification methods. Lawing recommended that keyworders have an English degree and a general knowledge of a broad range of topics. Hastings said that any indexer interested in becoming an archivist should have a history degree, librarian experience, and archivist training. He recommended the archivist program at Western Washington University, which "has the best training in the U.S." Full-text keywording "can be done by anyone," Pearlstein said, provided they have the right computer equipment. The drawback is that the equipment is very expensive.