new toys for SFX

There are two new important enhancements for SFX that have recently been released: RSI (Rapid Service Indicator) and bX.

RSI
RSI is the newest member of the SFX API family and is meant as a replacement of the existing JSI (Journal Subscription API). The API allows querying of the SFX database for items such as ISSN, ISBN, LCCN, OBJECT_ID, and CODEN adn presents them in an XML output. Some of the features of RSI include:

• RSI includes not only journals, but also books.
• RSI allows searching not just by ISSN, but also by OBJECT_ID, ISBN, LCCN, and CODEN.
• RSI has full support for consortia APIs. It also takes into account institute specific activation and allows the sending of IP information so that SFX can determine the appropriate institute.
• RSI can respond with YES/NO and also a third value "MAYBE" when multiple objects are found.

What RSI allows us to do in a real world scenario is embed SFX query results directly in a web page. This might include adding SFX information to the OPAC rather than have the user open a pop-up, or add SFX information to a VLE for direct linking to a resource.

bX
bX is ExLibris' scholarly recommender service and is available through the SFX menu or as a standalone API for embedding in other applications. The service is a lot like Amazon 'people who looked at this also looked at...' service. When a user performs a search and clicks on the SFX button, bX will recommend articles that it knows others have viewed as well as the original article the user searched for. For example if I searched for an article on MARC21 it might show me another ten articles that other users have viewed closely after they viewed the same MARC21 article.

The way bX does this is by taking usage logs from sites that implement bX and uses this data to determine relationships between articles. No personal data is transmitted, only data pertaining to which articles have been viewed in proximity to other articles.

This is a really big enhancement to SFX as rather than provide links based on information like author, title, year, etc; SFX can provide links based on usage patterns form thousands of users globally. This should provide for a far more 'human' approach to link recommendation rather than just the computational version that is currently available in SFX.