Thanks for the information. Looks like it is still being integrated. Do you know what state it is in? I am going to take a look at the contrib and see where it is at.
Thanks for your help, Xibin ryantxu wrote: > > check: > https://issues.apache.org/jira/browse/LUCENE-1387 > > On Oct 20, 2008, at 6:56 PM, xibin wrote: > >> >> Thanks for the reply. >> >> Let's say that I have a location code field in the document. Say >> there are >> 2000 indexed documents that have various location codes. When a user >> searches, she specifies her own location, as well as the maximum >> radius that >> she wants the distance to be, in addition to other search fields >> that she >> may specify. >> >> In order to satisfy this scenario, I am thinking about performing >> the search >> using all the other fields normally, but use a distance to dynamically >> filter out unwanted results. This way the distance calculation >> performs >> filtering, and the relevancy from other search fields does not >> change... >> >> Let me know if this makes sense to you. >> >> Thanks >> Xibin >> >> >> ryantxu wrote: >>> >>> I don't quite follow what you are trying to do.... >>> can you give a concrete example? >>> >>> What is missing from the Filter side? >>> >>> if you need to modify the lucene Query, a custom SearchComponent is >>> the way to go... >>> >>> >>> On Oct 20, 2008, at 6:13 PM, xibin wrote: >>> >>>> >>>> Hi - >>>> >>>> I have done some searching and haven't found what I was looking for. >>>> I hope >>>> this has not been discussed in the forum already. >>>> >>>> This is a question as well as looking for design ideas. >>>> >>>> I need the ability to dynamically filter out search results as they >>>> are >>>> being collected. The logic that I am developing cannot be statically >>>> applied >>>> at indexing time, so the data is NOT available in the indexed form. >>>> It can >>>> be derived/calculated using one or more of the indexed fields, and >>>> it's >>>> different for each query. The purpose of this "derived field" is to >>>> eliminate resulting indexes so only a subset is considered. This is >>>> very >>>> similar to the Filter concept already in the >>>> SolrIndexSearcher.QueryCommand. >>>> The difference is that I can't write a Lucene Query to obtain the >>>> subset of >>>> indices, I need to implement an algorithm involving using some of >>>> the fields >>>> in the documents. >>>> >>>> What I had in mind is the concept of a DyamicFilter. A DynamicFilter >>>> can be >>>> used in the HitCollectors (in SolrIndexSearcher) to perform dynamic >>>> filtering as results are being collected. DynamicFilters would be >>>> added into >>>> the SolrSearchIndexer.QueryCommand class so they can be called >>>> during >>>> collecting time. I considered writing a SearchComponent or a >>>> RequestHandler, >>>> and they seem to be a little bit off for the timing that I needed. >>>> >>>> The parameters to construct my DynamicFilters are passed in from >>>> http query >>>> params. I could pick them up and create the DynamicFilters in the >>>> QueryComponent as it creates a QueryCommand. The QueryCommand will >>>> then use >>>> it during Hit Collection in SolrIndexSearcher. >>>> >>>> I hope this captures the detail of what I am trying to do. I am >>>> looking for >>>> validation/alternative suggestions from an insider (yonik?). I feel >>>> bad >>>> having to do the intrusive modification here, and am open for >>>> suggestions. I >>>> would be interested in contributing this work if it turns out to be >>>> valuable. >>>> >>>> Thanks for reading. >>>> >>>> Xibin >>>> -- >>>> View this message in context: >>>> http://www.nabble.com/dynamic-filtering-tp20079841p20079841.html >>>> Sent from the Solr - Dev mailing list archive at Nabble.com. >>>> >>> >>> >>> >> >> -- >> View this message in context: >> http://www.nabble.com/dynamic-filtering-tp20079841p20080481.html >> Sent from the Solr - Dev mailing list archive at Nabble.com. >> > > > -- View this message in context: http://www.nabble.com/dynamic-filtering-tp20079841p20082651.html Sent from the Solr - Dev mailing list archive at Nabble.com.
