At 05:17 PM 6/4/2001 -0400, Matthew Schmidt wrote: >Lucene Developers and Users, > I have been assigned to look at the possibilty of using Lucene for > indexing a large amount of news articles and accessing the index in > through a search interface. Several things have drawn us to Lucene, > namely its speed and its price, but there are a few things we need > answered before we can go ahead with any implementation. Any help on the > following questions would be much appreciated. > >1) Does lucene support queries which include > or <. Eg. dateField < >some date or all id's > some id. Yes, as long as dates are stored in com.lucene.document.DateField format, you can use com.lucene.search.DateFilter to specify a date range in your query. > >2) Can lucene handle large lists of queries. Comparable to a SQL query of >where text="hi" in (some list of ids) I don't quite understand what you mean by a large list of queries. Do you mean a single query with a large number of conditions? or simultaneous processing of a large number of queries? Perhaps you can clarify more here? > >3) We need a way to order the index in a way so that the most recent >articles are listed first and therefore retrieved first. Because there is >the possiblity of several hundred thousand returned articles, this >ordering needs to take place when new articles are added to the index, not >on retrieval. Is this possible? Quite possibly. Someone has mentioned before on this mailing list that lucene search documents in chronological order that they were stored. However, I have not investigated this myself so you might want to perform some test to verify the validity of this claim. > >4) Finally, we need to short-circuit returns. Namely, if the user wants >just the 50 most recent articles, we need to just go get the first 50 and >then stop, not wasting processor time looking in the next million. I'm quite certain this is not built into lucene (Someone correct me if I am wrong). However you have the lucene source code. It shouldn't be too tough to add that piece of functionality. ;) > >If anyone can help me answer these, please let me know. > > Matthew Schmidt > <mailto:[EMAIL PROTECTED]>[EMAIL PROTECTED] > _______________________________________________ Lucene-users mailing list [EMAIL PROTECTED] http://lists.sourceforge.net/lists/listinfo/lucene-users
