Hello Oliver, We are very interested in group sorting based on some aggregation function also. Would you consider contributing your code to Lucene, or posting your results?
Tom Tom Burton-West Information Retrieval Programmer Digital Library Production Service University of Michigan Library http://www.hathitrust.org/blogs/large-scale-search On Mon, Feb 11, 2013 at 3:55 PM, Oliver Xu <oliver.xuy...@gmail.com> wrote: > Right. I understand that the TermFirstPassGroupingCollector is extended > from > Collector class too. Let me try. > > Many thanks to you! > > -----邮件原件----- > 发件人: martijn.is.h...@gmail.com [mailto:martijn.is.h...@gmail.com] 代表 > Martijn v Groningen > 发送时间: 2013年2月11日 19:29 > 收件人: java-user@lucene.apache.org > 主题: Re: About the Sorting of Groups during Grouping by > > Hi, > > My purpose is trying to sort the groups based on some way of aggregation > > values of the scores of all the documents in each group, such as their > sums > > or averages. > > > At the moment this isn't possible out of the box. You could implement your > own > collector that collects these statistics per group. Embed the the result in > a FieldComparator > that you use as custom sort in the TermFirstPassGroupingCollector. But this > is just a wild idea. > > > > > > > > > > There is an 'Enum TopGroups.ScoreMergeMode' in the > > org.apache.lucene.search.grouping package with the description as "How > the > > GroupDocs score (if any) should be merged." Is it exactly introduced for > > the > > purpose above? If yes, how to included it into the grouping code? A demo > > code would be perfect. > > > The enum is actually only used by the join module > (ToParentBlockJoinCollector). > The enum controls how the block join group scores are aggregated during a > distributed search. > > Martijn > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org > For additional commands, e-mail: java-user-h...@lucene.apache.org > >