On the boosting approach, you can have a mandatory field of title match and
optional match of userId with very high boost. This would have duplicates but
you don't need to do sorting to remove it. Just keep adding the result in the
order it comes and if you see that the title is already there in
Couple of approaches. (But not very sure if there are other better approaches)
1. Add a seperate field which is set to 1 or 0 depending upon if the self
description is present or not. Then you can search by this new field.
2. Along with each self-descrition add a common idenfier word. Then you ca
Hi Tom,
You are using a SimpleAnalyzer while indexing which will do some transformation
to your string before indexing. If you are using analyzer which does
transformation to the words during indexing you should use the same or similar
analyzer during querying as well.
Try the same with Keywor
My guess that this can happen when your document matches more than one
condition. For example first:25 could match lang:java as well??
- Original Message
From: hacklisp
To: java-user@lucene.apache.org
Sent: Thursday, May 21, 2009 10:03:52 AM
Subject: About sort questions
I search
Hi Eric LeVin,
I think whenever you reopen the indexReader you have to re-create indexSearcher
also. This is because reopen of indexReader will give you a new instance if the
underlying data is changed.
Function documentation for IndexReader.reopen
* If the index has not changed since this i
Hi,
I found the following piece of logic in DisjunctionMaxScorer.skipTo function
which may have a potential issue (marked in code as <<<>>>).
public boolean skipTo(int target) throws IOException {
if (firstTime) {
if (!more) return false;
heapify();
first
