Some users have reported that they can retrieve no full text when searching Newztext – and more often, Newztext Plus. In fact, the full text is there as always, its just that when the user uses the default sort order of “Score” smaller documents are more likely to be at the top of the results list – and the smallest are index records from Newzindex INNZ and NBR Index records.
Being small, but with subject indexing included, means that these documents will be scored as more relevant by the search server – i.e. higher density of searched words in the record. I would suggest that the search server is actually right! Especially since a human (well, a librarian or ex-librarian) has actually added the subjects after reading the article.
If the user sets the sort order to be “Date (earliest)” then they will get what appears to be only non-fulltext records as well – because the full text only starts around 1993, and the index records go back to 1840 or so.
If the user sets the sort order to be “Date (latest)” then they will see almost exclusively full text (from Newspapers and Newswires) – because they are added in their thousands every day, and will almost always be the latest (today’s date).
As you add more search terms, this effect is mitigated somewhat, as the relevancy scoring takes into account the presence of all or most of the terms as a higher relevancy than a small document with only one or a few of the search words.
If the user wants the full text only, then best to turn off the collections that are indexes – i.e.:
Newzindex, INNZ and NBR Index.
It is intended that we will add Faceted searching once all the changes are bedded down – in this case, a Facet for full text or indexes – so the user can just select the Facet after the results have been displayed.
Recent Comments