SRU Scan Data Model

["LeVan,Ralph" <levan@oclc.org>]

Title: SRU Scan Data Model

A first draft:

Search engines often create indexes on the fields that they search.  These indexes can consist of all or part of the contents of single fields or combinations of fields from records in their database.  Some of these indexing search engines are capable of exposing the lists of search terms that they have generated.  The SRU Scan Operation provides a mechanism for exposing those search terms.

The model is that the server has a sorted list of search terms for each scannable index.  The client provides a term that may be in the index to be scanned.  The server finds the actual search term from the index that is closest (closeness is not defined by the standard) and returns terms around that term based on the responsePosition and maximumTerms parameters provided by the client.  The responsePosition specifies the ordinal position that the nearest term should have in the list of terms returned. maximumTerms specifies the number of terms to be returned.  The server may return fewer terms.

If the responsePosition is zero or less, then the nearest term should not be returned in the response.  The first term returned in the response should be the term that comes N terms after the nearest term, where N=|responsePosition|+1;  (E.g., if responsePosition=-1, then the first returned term should be the second term following the nearest term.)

If the responsePosition if greater than the maximumTerms, then the closest terms should occur N terms after the last term returned, where N=responsePosition-maximumTerms.  (E.g., if responsePosition=11 and maximumTerms=10, then the last term returned should occur 1 position before the nearest term in the sorted index.)

Ralph