How fast does it need to be? I've done this sort of things for relevance evaluation with a driver in Python. Send the query, request 10 or 100 hits in JSON. Request only the URL field (fl parameter). Iterate through them until the URL matches. If it doesn't match, request more. Print the number.
Try it, I bet you would be surprised at how fast it is. You can run several copies of this script in parallel, maybe 10 or 20. Writing this in Solr seems like hitting a fly with a hammer. It is over-engineering. Build it in PHP first, even if you do want to do it in Solr. wunder On Jun 23, 2014, at 11:20 PM, Tri Cao <tm...@me.com> wrote: > It wouldn't be too hard to write a Solr Plugin that take a param docId > together with a query and return the position of that doc within the result > list for that query. You will still need to deal with the performance though. > For example, if the doc ranks at one millionth, the plugin still needs to get > at least 1M docs, and so the underlying collector still needs to sort through > 1M documents. > > Maybe your business requirement is different, but does it really make a > difference if a document ranks at one thousandth or one millionth? For > example with Google SEO, either you rank in the first 3 (precision at 3) or > you don't :) > > --Tri > > On Jun 23, 2014, at 10:55 PM, Jorge Luis Betancourt Gonzalez > <jlbetanco...@uci.cu> wrote: > >> Basically this is for analytical purposes, essentially we want to help >> people (which sites we’ve indexed in our app) to find out for which >> particular terms (in theory related with their domain) they are bad >> positioned in our index. Initially we’re starting with this basic “position >> per term” but the idea is to elaborate further in this direction. >> >> This logic por position finding could be abstracted effectively in a plugin >> inside Solr? I guess it would be more efficient to iterate (or fire the 2 >> queries) from within solr itself than in our app (written in PHP, so not so >> fast for some things) speeding up things? >> >> Regards, >> >> On Jun 24, 2014, at 1:42 AM, Aman Tandon <amantandon...@gmail.com > >> wrote: >> >> > Jorge, i don't think that solr provide this functionality, you >> have to >> > iterate and solr is very fast in this, you can create a script for >> that >> > which search for pattern(term) and parse(request) the records >> until get the >> > record of that desired url, i don't thing 1/3 seconds time to find >> out is >> > more. >> > >> > As per the search result analysis, there are very few people who >> request >> > for the second page for their query, otherwise mostly leave the >> search or >> > modify query string. So i better suggest you that the if the >> website has >> > the appropriate and good data it should come on first page, so its >> better >> > to come on first page rather than finding the position. >> > >> > With Regards >> > Aman Tandon >> > >> > >> > On Tue, Jun 24, 2014 at 10:35 AM, Jorge Luis Betancourt Gonzalez < >> > jlbetanco...@uci.cu > wrote: >> > >> > > Yes, but I’m looking for the position of the url field of >> interest in the >> > > response of solr. Solr matches the terms against the >> collection of >> > > documents and returns sorted list by score, what I’m >> trying to do is get >> > > the position of the a specific id in this sorted >> response. The response >> > > could be something like position: 5, or position 500. To >> do this manually >> > > suppose the response consists of a very large amount of >> documents >> > > (webpages) in this case I would need to iterate over the >> complete response >> > > to find the position, which in a worst case scenario >> could be in the last >> > > page for instance. For this particular use case I’m not >> so interested in >> > > the URL field per se but more on the position a certain >> url has in the full >> > > solr response. >> > > >> > > On Jun 24, 2014, at 12:31 AM, Walter Underwood >> <wun...@wunderwood.org > >> > > wrote: >> > > >> > > > Solr is designed to do exactly this very, very >> fast. So there isn't a >> > > faster way to do it. But you only need to fetch the URL >> field. You can >> > > ignore everything else. >> > > > >> > > > wunder >> > > > >> > > > On Jun 23, 2014, at 9:32 PM, Jorge Luis >> Betancourt Gonzalez < >> > > jlbetanco...@uci.cu > wrote: >> > > > >> > > > > Basically given a few search terms >> (query) the idea is to know given >> > > one or more terms in which position your website is >> located for those >> > > specific terms. >> > > > > >> > > > > On Jun 24, 2014, at 12:12 AM, Aman >> Tandon <amantandon...@gmail.com > >> > > wrote: >> > > > > >> > > > > > What kind of search criteria, >> could you please explain >> > > > > > >> > > > > > With Regards >> > > > > > Aman Tandon >> > > > > > >> > > > > > >> > > > > > On Tue, Jun 24, 2014 at 4:30 >> AM, Jorge Luis Betancourt Gonzalez < >> > > > > > jlbetanco...@uci.cu > >> wrote: >> > > > > > >> > > > > > > I’m using Solr for an >> analytic use case, one of the requirements is >> > > > > > > basically given a >> search query get the position of the first hit. I’m >> > > > > > > indexing web pages, >> so given a search criteria the client want’s to >> > > know >> > > > > > > the position (first >> occurrence) of his webpage in the result set (if >> > > it >> > > > > > > appears at all). Is >> any way of getting this position without >> > > iterating and >> > > > > > > manually checking the >> solr response? >> > > > > > > >> > > > > > > Greetings, >> > > > > > > >> > > > > > > >> > > > > > > VII Escuela >> Internacional de Verano en la UCI del 30 de junio al 11 de >> > > > > > > julio de 2014. Ver >> www.uci.cu >> > > > > > > >> > > > > >> > > > > VII Escuela Internacional de Verano en >> la UCI del 30 de junio al 11 de >> > > julio de 2014. Ver www.uci.cu >> > > > >> > > > >> > > > >> > > > >> > > >> > > VII Escuela Internacional de Verano en la UCI del 30 de >> junio al 11 de >> > > julio de 2014. Ver www.uci.cu >> > > >> >> VII Escuela Internacional de Verano en la UCI del 30 de junio al 11 de julio >> de 2014. Ver www.uci.cu
