Right, so you're using edismax? This is expected. You can do
a number of things:
1> change the parameters of edismax
2> have your app filter out returns that dive beneath some threshold
that is relative to the score of the first doc in the list.
But I don't see why, given your example, it makes a difference. Because
you got the document you were looking for in this instance says nothing
abou the next time when you search on something different....
Best
Erick
On Mon, Aug 15, 2011 at 2:22 PM, Donald J. Organ IV
<dor...@donaldorgan.com> wrote:
> OK I am doing a search using the following fields name^2.0 code^1.8
> cat_search^1.5 description^0.8
>
> I am searching for: free range dog nips
>
> I am getting back 2 documents the first is the document I am looking for, and
> contains those works in the name field, as the name field is "Free Range Dog
> Nip Chicken Breast Wraps"
>
>
> The second looks like its matching because those words are contained within
> the description.
>
>
>
> ----- Original Message -----
> From: "simon" <mtnes...@gmail.com>
> To: solr-user@lucene.apache.org
> Sent: Monday, August 15, 2011 1:59:17 PM
> Subject: Re: Minimum score filter
>
> The absolute value of a relevance score doesn't have a lot of meaning and
> the range of scores can vary a lot depending on any boost you may apply.
> Even if you normalize them (say on a 1-100 scale where 100 is the max
> relevance) you can't really draw any valid conclusions from those values.
>
> It would help if you described exactly what problem you're trying to solve.
>
> -Simon
>
> On Mon, Aug 15, 2011 at 1:02 PM, Donald J. Organ IV
>
