-
Suggestion
-
Resolution: Won't Fix
NOTE: This suggestion is for Confluence Cloud. Using Confluence Server? See the corresponding suggestion.
Right now, our searches are complicated by the fact that each field on the object goes into a different term in the index.
The advantages of doing this are that we can weight titles higher, for example.
The disadvantage is that if someone searches for "foo AND bar", you'll only get a match if foo and bar are in the same field: an email addressed to Charles, with beer in the message body won't match "charles AND beer"
I've fixed this for the simplest case - title of a CEO for CONF-3426, but we need a more general solution where everything we want to be searched on gets put in a single field.
- is related to
-
- Closed
[AI-648] Make one big searchableText term in Lucene containing all searchable text for that entity
Compose all searchable fields in one "total" is generally a good idea, but anyway it makes sense to build more complex boolean Lucene query with weights for different fields, for example
BooleanQuery resultQuery = new BooleanQuery(); resultQuery.add(new MatchAllDocsQuery(), Occur.MUST); fieldsToSearch = new String[]{"title", "total"}; boostOptions = new float[]{1.0000f, 0.0000f}; fieldOptions = new BooleanClause.Occur[]{Occur.SHOULD, Occur.MUST}; for (int i = 0; i < fieldsToSearch.length; i++) { Query fieldQuery; try { SolrQueryParser queryParser = new SolrQueryParser(schema, fieldsToSearch[i]); queryParser.setDefaultOperator(SolrQueryParser.AND_OPERATOR); fieldQuery = queryParser.parse(keywords); fieldQuery.setBoost(boostOptions[i]); resultQuery.add(fieldQuery, fieldOptions[i]); } catch (Exception e) { log.debug("Error during query parsing\n" + e.getStackTrace()); } }
Probably you can also check standard MultiFieldQueryParser
http://lucene.apache.org/java/2_3_2/api/org/apache/lucene/queryParser/MultiFieldQueryParser.html
Alexey Alexey
added a comment - - edited Compose all searchable fields in one "total" is generally a good idea, but anyway it makes sense to build more complex boolean Lucene query with weights for different fields, for example
BooleanQuery resultQuery = new BooleanQuery();
resultQuery.add( new MatchAllDocsQuery(), Occur.MUST);
fieldsToSearch = new String []{ "title" , "total" };
boostOptions = new float []{1.0000f, 0.0000f};
fieldOptions = new BooleanClause.Occur[]{Occur.SHOULD, Occur.MUST};
for ( int i = 0; i < fieldsToSearch.length; i++) {
Query fieldQuery;
try {
SolrQueryParser queryParser = new SolrQueryParser(schema, fieldsToSearch[i]);
queryParser.setDefaultOperator(SolrQueryParser.AND_OPERATOR);
fieldQuery = queryParser.parse(keywords);
fieldQuery.setBoost(boostOptions[i]);
resultQuery.add(fieldQuery, fieldOptions[i]);
}
catch (Exception e) {
log.debug( "Error during query parsing\n" + e.getStackTrace());
}
}
Probably you can also check standard MultiFieldQueryParser
http://lucene.apache.org/java/2_3_2/api/org/apache/lucene/queryParser/MultiFieldQueryParser.html
Given that this issue has been open for several years with no votes and no traction, marking as "Won't Fix".