Matrix, Search Engines: WebCrawler

Key Links

Organization

• WebCrawler is an exclusively searchable database of Web documents, built on a custom software engine written by the author using C.
• Features and Limitations:
  • Supports simple Boolean OR (by default) or Boolean AND (by clicking the Forms checkbox) across multiple keywords, but doesn't handle Boolean Not, complex Boolean combinations, or Proximity Searching.
  • The databases creates its indexes by identifying words on space and punctation boundaries, converts them to lowercase, and strips off common suffixes such as -s, -er, and -ment. It also filters out common words such as web, Internet, be, and, and or.
  • The server weights the hits on the quality the match between keywords and documents, then returns the highest ranking documents in sorted order. The user specifies the number of hits as discrete amounts (10, 25, 100, or 500).
  • The engine indexes and searches across filenames, document titles, as well as full textual content.
• WebCrawler provides both Forms and Non-forms interfaces to the search engine, however Forms support is required for most of the search features.
• The information catalogued by WebCrawler has no specific focus or content restrictions.

Administration

• Document information is gathered automatically by a custom Web searcher and from user-suggested URL's.
• Average response time for basic access is about 5 seconds, and searches return within 30 seconds.
• The server runs on a Pentium computer under NextStep, and the dcoument gathering engine operates from a similar second machine. The WebCrawler index currently contains information on over 420,000 documents.
• The layout and organization of the server are very simple and the information provided is quite helpful. The flexibility of the search engine (smart truncation, etc), the simplicity of the search page, and the formatting of the search results make the server ideal for new and experienced users.
• Additional Services
  • The help page demonstrates sample queries, with suggestions for improving search quality, and a description of the indexing process.
  • The server maintains a list of the Top 25 URLs linked from other documents. This is not a reflection of the actual traffic on a particular document, but the number of hotlists and index pages that include a pointer to it.
  • The server allows users to suggest documents for inclusion into the search.