Magazine article American Libraries

Technically Speaking: Exhibitors Uniting in Their Desire to Provide Federated Searching

Magazine article American Libraries

Technically Speaking: Exhibitors Uniting in Their Desire to Provide Federated Searching

Article excerpt

The hot new phrase at this year's ALA Annual Conference exhibits was "federated searching." The use of the phrase came out of the Open Archives Initiative's Protocol for Metadata Harvesting (OAI-PMH). Using the protocol, a single server "harvests" metadata records from multiple remote federated databases. The resulting centralized metadata file is in turn searchable by researchers. But the phrase was quickly expanded to refer to any defined set of multiple remote resources that are searched simultaneously, regardless of whether their metadata records are harvested and put into a single server, or left in the remote databases and targeted by clients doing distributed searching. Vendors who used the term in Toronto were almost exclusively referring to distributed searching (also called broadcast searching) of multiple remote databases by Z39.50 and native interfaces.

Kate Noerr, CEO of MuseGlobal, which produces MuseSearch and a whole family of Muse products aimed at the resource retrieval integration market, believes that federated searching itself is becoming a commodity item with little differentiation among vendor offerings because the technology is fairly straightforward and is being widely adopted. One of the associated services that is not so straightforward, she asserts, is maintaining the links and updating the searching software to reflect changes in search interfaces and functionality in the thousands of information servers that a vendor of federated search services must link to.

Another area of federated search services in which vendors can distinguish themselves, she adds, is post-retrieval processing. I got the sense that all the vendors of federated search services agree that some aspects of post-retrieval processing, such as deleting duplicate results, or "deduplicating," and sorting results in multiple ways, are tough nuts to crack. WebFeat, which markets a federated search engine and supporting technologies, does not even try to deduplicate results or to sort in multiple ways across databases of origin.

Putting ZING into Searching

The increasing prominence of federated searching technology is also influencing the development of new library catalog search tools. Dynix is hitching its technology wagon to one of the new Z39.50 International: Next Generation (ZING) protocols. Search/Retrieve Web (SRW) service, one aspect of the ZING suite of protocols, marries Z39.50 semantics with the Extensible Markup Language (XML) to create a web service that can utilize many of the new web service standards such as SOAP (Simple Object Access Protocol), clients such as Microsoft's.Net, and tools such as Javascript. "It blows away all the other search technologies we have investigated," says Jim Wilson, one of the founders of Dynix, who is back with the company as senior library advisor, primarily to enhance Dynix's standing with academic libraries through the adoption of the latest technologies.

The company has also decided to implement SRW natively in its Horizon system. This is a decision that could have a positive impact on searching standardization. For years, the semantics of the Z39.50 search and retrieve protocol have been a potential, but almost unutilized, tool for standardizing catalog searching among all Library Management Systems. At present, VTLS's Virtua is the only LMS that utilizes Z39.50 search functionality natively. By deciding to combine the Z39.50 semantics and XML markup, Dynix is taking a giant leap into search standardization. Like the VTLS decision a few years ago, this move by Dynix represents a challenge to the entire LMS market. Let's hope the other vendors rise to the occasion.

In casting around for the best search engine to implement the Search/Retrieve Web Service, Dynix is evaluating Lucene (or Jakarta Lucene, as it is also known), an open source "high-performance, full-featured text search engine written entirely in Java," according to its developers (jakarta. …

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


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.