1. Use edismax and the boost parameter. Multiplicative boosting works much, much better than the additive boost used in dismax.
2. Use a log-scaled popularity value. If you know popularity is always greater than 1, use “1 + log(popularity)”. It is a bit safer to use “1 + log(max(1, popularity))”. 3. Multiple that value by a weight to make it a tie-breaker. If two items have the same title (or text), you want the more popular one first. But you rarely want a different result boosted above a match. No matter how popular the movie “Twilight” is, it should not rank above “Twilight Zone” for the query “twilight zone”. wunder Walter Underwood wun...@wunderwood.org http://observer.wunderwood.org/ (my blog) > On Sep 18, 2016, at 11:49 PM, Rajendra Gaikwad <rajendra...@gmail.com> wrote: > > Hi Midas, > I am sure,even though you have results with constant relavance, sorting > will give you expected results. > Anyway there is parameter bf in dismax parser. Add parameter > bf=field(popularity) to query, It is similar to boost items by value in > popularity field. hence ultimately higher popularity items occurs on top. > Add parameter to your query bf=field(popularity)&defType=edismax > > > Thanks, > Rajendra Gaikwad > > On Mon, Sep 19, 2016, 11:55 AM Midas A <test.mi...@gmail.com> wrote: > >> my use case do not suggest me to sort . i have set of data with same >> relevance. >> >> what should be query in that case . >> >> On Mon, Sep 19, 2016 at 11:51 AM, Rajendra Gaikwad <rajendra...@gmail.com> >> wrote: >> >>> Hi Midas, >>> >>> Sort search results on popularity field by desc order. >>> E.g popularity is field in the index which stores popularity information. >>> >>> http://localhost:8983/solr/mycollection/select?q=*:*&sort=popularity >> desc >>> >>> Thanks, >>> Rajendra Gaikwad >>> Please execuse typo >>> >>> >>> >>> On Mon, Sep 19, 2016, 11:36 AM Midas A <test.mi...@gmail.com> wrote: >>> >>>> i have n items in my search result with popularity (1,2,3,4....n) . I >>> want >>>> higher popularity item should come first then next popularity item >>>> >>>> >>>> say for example >>>> a) item with popularity n, >>>> b) item with popularity n -1, >>>> c) item with popularity n -2, >>>> d) item with popularity n - 3, >>>> e) item with popularity n - 4, >>>> f) item with popularity n - 5, >>>> .... >>>> .... >>>> y) item with popularity 2, >>>> z) item with popularity 1, >>>> >>>> >>>> what should be my query if relevance for items are constant >>>> >>> -- >>> >>> sent from mobile, execuse typo >>> >> > -- > > sent from mobile, execuse typo
