You're not going to be able to look at field boosts by themselves to judge relevancy because it's very much a data-driven optimization problem. For example, if you only sell iPhone cases but no iPhones, a search for "black iphone" should show a bunch of black iPhone cases at the top of the results. But if you do sell iPhones themselves, you'll likely see them rank low in the results because they typically have names like "Apple iPhone 6s Plus 64 GB - Black" and your cases just have "iPhone Case - Black". More of the search terms match the shorter field value and so it scores better.
Approach the problem methodically and collect data. There are several evaluation metrics that will not only help you quantify the problem but also gauge how much your tuning efforts have improved things. MRR and DCGS are good places to start. https://en.wikipedia.org/wiki/Category:Information_retrieval_evaluation Also take a look at Quepid (full disclosure: my company makes it). It'll let the business folks rank the results for searches and you'll be able to do search regression tests against those judgement lists as you tweak things. k/r, Scott On Thu, Mar 17, 2016 at 4:36 AM, Robert Brown <r...@intelcompute.com> wrote: > Hi, > > I currently have an index of ~50m docs representing shopping products: > name, description, brand, category, etc. > > Our "qf" is currently setup as: > > name^5 > brand^2 > category^3 > merchant^2 > description^1 > > mm: 100% > ps: 5 > > I'm getting complaints from the business concerning relevancy, and was > hoping to get some constructive ideas/thoughts on whether these boosts look > semi-sensible or not, I think they were put in place pretty much at random. > > I know it's going to be a case of rounds upon rounds of testing, but maybe > there's a good starting point that will save me some time? > > My initial thoughts right now are to actually just search on the name > field, and maybe the brand (for things like "Apple Ipod"). > > Has anyone got a similar setup that could share some direction? > > Many Thanks, > Rob > > -- Scott Stults | Founder & Solutions Architect | OpenSource Connections, LLC | 434.409.2780 http://www.opensourceconnections.com
