Magazine article Online

Vivísimo, Central Search, TIME Magazine, and the Open Directory Project

Magazine article Online

Vivísimo, Central Search, TIME Magazine, and the Open Directory Project

Article excerpt

The first pick is a powerhouse combo of a federated search engine enhanced by Vivísimo's clustering technology in the new release of Serials Solutions' Central Search. This will make most library patrons appreciate the quality and variety of information and the guidance on offerings from their libraries. The other pick is the TIME magazine archive with more than 320,000 articles covering all 84 years of the news magazine-a very convenient and reliable ready-reference source, free for subscribers of the print edition. The pan is the Open Directory Project, which has entries that have become so outdated and misinforming that it triggered the creation and use of the NOODP metatag to alert Web search engine crawlers to use the metadata of the site instead of the ODP description in their results lists.

VIVÍSIMO AND CENTRAL SEARCH

Central Search is Serial Solutions' federated search engine-not the typical database I usually review in this column. I cover it here because it brings the best out of the mix of the abstracting and indexing (A&I) and full-text databases, journal archives, digital encyclopedias, and ready-reference sources to which librarians subscribe or-in cases of open access sources-provide access to. Through its Article Finder link resolver, Central Search also enhances the results lists of A&I records with links to the primary documents-which are under the patrons' noses, but people remain oblivious to them. Central Search was a smart and pretty federated search engine. Since its enhancement by Vivisimo's clustering engine in October, it has become a very smart and very pretty resource discovery tool.

Many other information services have been doing result clustering to provide at-a-glance clues about the composition of the results list (typically from a single database) by topics, keywords, source publications, authors, document types, or ages. However, it's not to the tune of hundreds of databases from different services, and it's typically by a single criterion rather than by a number of data elements at one fell swoop.

The entries in the clusters (along with the information posted) provide hints for drastically and instantly reducing the results set to a manageable ad hoc temporary subset by user-chosen topical criteria or other angles, such as journals or authors listed in the cluster sidebar.

The entries, being from multiple sources, can be redundant and inconsistent. History shows that they are often that way even within a single database. The heterogeneous mix of databases from different content providers increases the inconsistency. The cluster provides an indirect warning on what a Sisyphean task it would be to do a cross-database search by descriptors, authors, or journal names-and it offers a bypass solution to the problem. It also gives a sense of the overlap among databases.

If check boxes were displayed instead of the square bullet gizmos, users could consolidate the spelling variants of journal names such as the ones for IEEE Transactions on Power Electronics. For the journal cluster, Serials Solutions is in the best position, automatically and instantly, to do much of the consolidation using the many journal name variants in its A-Z source list (which is more like a knowledgebase). There are no "InstaSolutions" for the author names and descriptor variants scattered around in the cluster list, but it is an acceptable compromise in exchange for the immense power offered even by the not exactly squeaky-clean cluster terms.

This is exactly the type of application that would make users of the general Web search engines (which serve up increasingly polluted results from an unpredictable medley of sources) appreciate their libraries and understand that the most efficient way of finding pertinent information is to use the professional databases available with their free library cards. For libraries, the yearly cost for 100 databases with a consortial discount is less than $200 a pop. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.