Web Site Searching and the User Experience
BayCHI, January 1999
Avi Rappoport, Search Tools Consulting

Search Results List Interface

• More information is usually better

• Standard Results

• Page Title (linked to page)

• Meta Description, if available, otherwise page summary

• Date (publication date rather than mod date if possible)

• File Size (especially for large files)

• URL

• Examples:

  • HyperMall (Xavatoria) - good results with a simple Perl CGI

  • US Patent Office (ISearch) - rich results with lots of metadata

• Special Information for Special Cases

  • Formatted Text such as poems, schedules, acquired data, invoices

  • Non-HTML files: PDF, word processing, spreadsheets

  • Examples:

    • Shakespeare's Plays (WebGlimpse) - shows context

    • PBS (Magnifi) - shows thumbnail of file (particularly good for media searching)

• Relevance Ranking

  • Order of results

  • Rank Indicator: percent, filled bars etc.

  • Examples

    • NorthcoastWeb.com (Microsoft Index Server) - "Score" percent is misleading: what does it mean? Also focusses on file name rather than titles (which can be very long).

    • UK Government Information Service (Muscat) - Score is more obvious, results shows matched terms, but doesn't remove duplicates!

    • DeAnza College (Apple e.g., aka Sherlock) - ranks according to internal algorithm, not all terms must be matched.

Table of Contents | Previous Page | Next Page