I remember when FAST (when it was still FAST) came to our enterprise to pitch 
their search when we were looking to replace our alta vista search engine with 
*something* and they demonstrated that relevance tool for business side. While 
that thing was awesome, I've never seen anything close to it in the solr world 
where I ended up going instead of the soon to be doomed FAST search. Also that 
tool was totally manual and of limited use in a very large corpus/catalog. Sort 
of like just applying a bandaid to a larger problem.


Splainer will only detail the reasons things show up in one query but won't 
solve a bigger relevancy problem. On the other hand, there are several ways to 
skin this cat. There are solutions which analyze logs for outlying cases and 
feed back into solr these results to automatically improve relevancy. I don't 
think most any of these are open source and some are quite proprietary.


If your company could afford to assign a buisdev guy to tweeking individual 
searches, I'm sure they could instead get some jr devs to go over query logs 
inspecting outlying cases like zero results/too many results then look at if it 
is a data issue or a query issue. And then recommend changes in the appropriate 
domain.


Thanks

Robi

________________________________
From: Charlie Hull <char...@flax.co.uk>
Sent: Thursday, December 14, 2017 1:24:42 AM
To: solr-user@lucene.apache.org
Subject: Re: Any Insights SOLR Rank tuning tool

On 13/12/2017 20:18, Sharma, Abhinav wrote:
> Hello Folks,
>
> Currently, we are running FAST ESP as a Search System & are looking to 
> migrate from FAST ESP to SOLR.
> I was just wondering if you Guys have any built-in Relevancy tool for the 
> Business Folks like what we have in FAST called SBC (Search Business Center)?
>
> Thanks, Abhi
>
I'd second Quepid as we've used it for several projects where migration
is an issue (disclaimer: we're partners with OSC and resell Quepid).

Migration is a tricky thing to get right: the business side want the new
engine to behave like the old one, but don't understand the technical
issues when you're putting in a totally different core engine; technical
folks don't necessarily understand the business drivers behind making
the transition as painless as possible for users. Developing tests (and
being able to compare both sets of search results) is essential.
Remember that you might even have to replicate some 'wrong' behaviour of
the old engine as people are used to it!

Cheers

Charlie

--
Charlie Hull
Flax - Open Source Enterprise Search

tel/fax: +44 (0)8700 118334
mobile:  +44 (0)7767 825828
web: www.flax.co.uk<http://www.flax.co.uk>

________________________________

This communication is confidential. Frontier only sends and receives email on 
the basis of the terms set out at http://www.frontier.com/email_disclaimer.

Reply via email to