Facet ranges in SRU 2.0

["Edo Plantinga" <Edo.Plantinga@ictu.nl>]

Title: Re: Facet ranges

Hi all,

 

As a new member of this mailing list, please allow me to introduce myself. My name is Edo Plantinga and I work for the Dutch  government. We are  currently migrating the search engine behind the national Dutch government website overheid.nl ("government.nl") to a new search engine.   We are taking this opportunity to make the indexed data available  via an SRU webservice.

 

Before I start with my questions: my complements for the standard that you all are maintaining, keep up the good work!

 

I have asked Ray some questions about faceted search and he suggested that I post them to this forum:

 

=======

We would like to have range / date facets and present them to the user as "Published last week", "Published last month" etc. on which the user can drill down. Naturally, these facets are a moving window (i.e. the results tomorrow will be different from today).

Taking a closer look at the  definition of facet ranges, I cannot find a way to label the groups that you  can refine on. For example, for our collection it is common to define a date  facet range with labels like 'Published last week', 'Published last month'. I can see how you would *define* these ranges, but not how to *label* them. Am I  missing a something?
=======

Ray's answer: 

Edo -  I'm not entirely sure what you mean by facet labeling.  For the facet range feature, the client declares that for this particular request a particular facet is to be treated as a "range" facet, and then further specify an applicable range.  So for example (as the example in the draft spec) the client declares that dateOfPublication is to be treated as a range facet, with applicable range 20010101 through 20040404, meaning that counts should be reported only when dateOfPublication is between those specified dates, even though there may be dates outside of that range.

 

But the example you gave is quite different. In your example, the range is server defined (as opposed to client defined). You might have an index of dates "publishedLastWeek", publishedTwoWeeksAgo", and so on. The index publishedLastWeek would (presumably) contain only dates from last week, and so there would be no need to indicate any constraints by facet range limits - that is, if I understand this correctly.  And so these could perhaps be treated as normal facets which do not need further refinement. 

  

I would be happy for us to explore this further. The real expert on facets within the committee is Ralph LeVan. 

=======  

My response to this: Ok, now I see that the *client* can request a facet. However, I still cannot see how you would define the <facetDisplayLabel> (ie the text that is displayed to the user) in the request parameters. I do see the <facetDisplayLabel> in the response XML, so where is that defined? Server-side? If so, is there also a way to define something like <rangeFacetLabel>Last week</rangeFacetLabel> in the request parameters?

 

Does anyone have any thoughts on this? Maybe some facet requests and responses would help clarify things for me.

 

Kind regards,

 

Edo