Woogle4MediaWiki/Decider manual

From TeamWeaverWiki

< Woogle4MediaWiki
Revision as of 11:18, 28 April 2010 by Happel (Talk | contribs)
(diff) ← Older revision | Current revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Contents

Capabilities

See the start page for links to slidecasts and an online demo.

Features

  • Searching the Wiki and external content
    • Keyword-search with wildcards, phrases etc. (Lucene syntax)
    • Search external data from within MediaWiki (in WoogleRemote using Integrated search backend)
    • Autocomplete of search results
    • Seamless integration in MediaWiki (replaces built-in search; can be disabled by individual users via MediaWiki preferences)
    • Social search (Describe and discuss queries, (Social ranking))
    • Searches all namespaces
  • Need guidance
    • Popups for "Red links" provide meta-information on need
    • Statistics of unsatisfied queries
    • ...
  • Convenience features
    • Supports different languages (en and de supplied by default, further languages can easily be supported)
    • Native PHP extension (in WoogleNative mode)

Limitations

  • Accesses all pages in all namespaces - currently no configuration possible (but can be relatively easy changed in the code)
  • Does not respect any read restrictions/access rights on MediaWiki content (certain access control is possible in WoogleRemote using an Integrated search backend)
  • For minor limitions of specific releace also consult the Woogle4MediaWiki/Release Notes

Comparisons

Options for search in MediaWiki are described on Woogle's MediaWiki.org page and at http://www.wiki4enterprise.org/index.php/Suche_in_Wikis (in German).

Performance

  • WoogleRemote
    • Integrated Search is the core determinant for indexing and query performance - it can typically handle large amounts of data
    • Communication and even network latency between the PHP/MediaWiki frontend and the Java/TeamWeaverIS backend is not an issue to our experience
  • WoogleNative
    • On state-of-the-art hardware, we were able to work with a 50,000 pages subset of Wikipedia, serving queries well below one second
    • However, we did not yet test this scale with massive requests
    • We recommend to restrict search-term autocomplete to title matches only for Wikis larger than 10,000 pages for performance reasons