CERN Accelerating science

Previous:		Dublin Core Metadata	(See printing version)
Next:		CERN Search Syntax Summary

The Search Engine used at CERN to perform search across all CERN Web servers is UltraSeek provided by the company Infoseek (see the article "CERN Search Syntax Summary" in this CNL). Information on searches on a sub-set of the CERN Intranet both from the reader's and authors' point of view is provided in the corresponding WebOffice Web page at URL: http://www.cern.ch/WebOffice/Doc/Search/.

Since December 7th, CERN is running UltraSeek server 3.02. This article is listing the Release Notes (new features and implementation changes).

New features:

• Date range searches.

Users can specify a range of document dates in their searches.
 
• New languages support.

Lexical support has been added for Swedish, Danish, Finnish and Norwegian.
UltraSeek supported only English, Dutch, French, German, Italian, Portuguese
and Spanish in previous version.
 
• Cooperative spidering.

There is now support enabling cooperation between multiple spidering instances
of UltraSeek (they are able to feed hyperlink URLs to each other).
 
• Merge collection.

A new kind of collection can be built by merging the indices from other collections.
This is useful if you have many collections and you want to produce a single set
of indices for search.
 
• XML support.

Documents written in the Extensible Markup Language (XML) can be indexed.
 
• SSL support.

The spider supports the Secure Sockets Layer and can fetch documents using HTTPS.
Available as an option.
 
• HTML meta tag names.

This feature allows to specify exactly which "meta" tags to use for the purposes of title
override, summary override, date override, keywords and publisher.
 
• Word spam detection thresholds.

Ultraseek indexer includes an automatic spam detection algorithm. It decides that
a document is spamming a certain word if that word occurs more often than a certain
number of times within a sequence of 100 words.
 
• Disallow rext or links from apparent directory listing documents.

You can disable the indexing of text or the following of links from documents that are
directory listings.
 
• Document type parsing.

You can customize the names of document content-types.
 
• Add URL collection determination.

UltraSeek Server can add the URL to all collections that allow it, and can also redirect
the request to another instance of UltraSeek Server if cooperative spidering is configured.
 
• URL status.

This feature can automatically determine which collection a URL belongs to.
 
• Revisit site.

You can request that all URLs from specific site should be revisited.
 
• Site and URL listings.

You can get a listing of all known URLs on specified site. Site and URL listings are now
available to non-administrative users.

Implementation changes:

• Filter Lotus Domino navigation links

Filter Lotus Domino Navigation Links feature has been improved.
 
• Add URL speed of indexing.

URLs added through the "Add URL" interface, during times that the spider is busy, get indexed
more quickly than did in previous version.
 
• Use HTTP Keep Alives.

A bug in the spider that would cause it to timeout when communicating with some HTTP servers
with keep-alives turned on has been fixed.
 

For comments send e-mail to www.support@cern.ch

For matters related to this article please contact the author.