I am looking for a way of coding the lucene fuzzy query that searches all the documents, which are relevant to an exact phrase. If I search "mosa employee appreciata", a document contains "most employees appreciate" will be returned as the result.
I tried to use:
FuzzyQeury = new FuzzyQuery(new Term("contents","mosa employee appreicata"))
Unfortunately, it empirically doesn't work. The FuzzyQuery employs the editor distance, theoretically, "mosa employee appreciata" should be matched with "most employees appreciate" provide the appropriate distance is given. It seems a bit odd.
Any clues? Thank you.
There are two likely problems here. First: I'm guessing the "contents" field is being analyzed such that "most employees apreciate" is not a term, but rather three terms. Defining as a single term is not appropriate in this case.
However, even if the content listed is a single term, a second likely problem we have is that there is too much distance between the terms to get a match. The Damerau-Levenshtein distance between mosa employee appreicata and most employees appreciate is 4 (the approximate distance, incidentally, between my average first shot at spelling
"Damerau-Levenshtein" and the correct spelling). Fuzzy Query, as of 4.0, handles edit distances of no more than 2, due to performance constraints, and the assumption that larger distances are usually not particularly relevant.
If you need to perform a phrase query with fuzzy terms, you should look into either MultiPhraseQuery, or combine a set of SpanQueries (especially SpanMultiTermQueryWrapper and SpanNearQuery) to meet your needs.
SpanQuery[] clauses = new SpanQuery[3];
clauses[0] = new SpanMultiTermQueryWrapper(new FuzzyQuery(new Term("contents", "mosa")));
clauses[1] = new SpanMultiTermQueryWrapper(new FuzzyQuery(new Term("contents", "employee")));
clauses[2] = new SpanMultiTermQueryWrapper(new FuzzyQuery(new Term("contents", "appreicata")));
SpanNearQuery query = new SpanNearQuery(clauses, 0, true)
And since none of the individual terms have an edit distance greater than 2, this should be more effective.
ComplexPhraseQueryParser handles fuzzy searching on phrase words - i.e., specify the words that should be fuzzy searched and those that should not. Works as follows
Query query = new ComplexPhraseQueryParser("content", analyzer)
.parse("some test~ query~ blah blah");
Seems to work nicely. Not sure about performance, however but seems to work well on small data sets.
I had some (very small) millage with the following:
String[] searchTerms = searchString.split(" ");
FuzzyLikeThisQuery fltw = new FuzzyLikeThisQuery(searchTerms.length, new StandardAnalyzer());
Arrays.stream(searchTerms)
.forEach(term -> fltq.addTerms(term, FIELD, SIMILARITY_IN_EDITS, PREFIX_LENGTH);
This query matches far too distant strings with the index. String that don't match are ones where each of the terms are distant by more than 2 edits from the terms used in the indexed content.
Please use at your own peril.
The answer from femtoRgon is great! Thank you.
There is another way to solve this problem.
//declare a mutilphrasequery
MultiPhraseQuery childrenInOrder = new MultiPhraseQuery();
//user fuzzytermenum to enumerate your query string
FuzzyTermEnum fuzzyEnumeratedTerms1 = new FuzzyTermEnum(reader, new Term(searchField,"mosa"));
FuzzyTermEnum fuzzyEnumeratedTerms2 = new FuzzyTermEnum(reader, new Term(searchField,"employee"));
FuzzyTermEnum fuzzyEnumeratedTerms3 = new FuzzyTermEnum(reader, new Term(searchField,"appreicata"));
//this basically pull out the possbile terms from the index
Term termHolder1 = fuzzyEnumeratedTerms1.term();
Term termHolder2 = fuzzyEnumeratedTerms2.term();
Term termHolder3 = fuzzyEnumeratedTerms3.term();
//put the possible terms into multiphrasequery
if (termHolder1==null){
childrenInOrder.add(new Term(searchField,"mosa"));
}else{
childrenInOrder.add(fuzzyEnumeratedTerms1.term());
}
if (termHolder2==null){
childrenInOrder.add(new Term(searchField,"employee"));
}else{
childrenInOrder.add(fuzzyEnumeratedTerms2.term());
}
if (termHolder3==null){
childrenInOrder.add(new Term(searchField,"appreicata"));
}else{
childrenInOrder.add(fuzzyEnumeratedTerms3.term());
}
//close it - it is important to close it
fuzzyEnumeratedTerms1.close();
fuzzyEnumeratedTerms2.close();
fuzzyEnumeratedTerms3.close();
Related
Is it possible to find the position of words with a match when the indexed field isn't stored?
for example:
Query: "fox over dog"
Indexed text of matched doc: "The quick brown fox jumps over the lazy dog"
What I want: [4,6,9]
Note1: I know text can be highlighted using Lucene but I want the position of the words
Note2: The field isn't set to be stored by Lucene**
I have not done this for practical purposes - just to give a pseudo code and pointers that you can experiment with to reach to correct solution.
Also, you have not specified your Lucene version, I am using Lucene 6.0.0 with Java.
1.While Indexing, set these two booleans for your specific field for which positions are desired. Lucene will be able to give that data if indexing has stored that information otherwise not.
FieldType txtFieldType = new FieldType(
TextField.TYPE_NOT_STORED);
txtFieldType.setStoreTermVectors(true);
txtFieldType.setStoreTermVectorPositions(true);
2.At your searcher, you need to use Terms , TermsEnum & PostingsEnum like below,
`Terms terms = searcher.getIndexReader().getTermVector(hit.doc, "TEXT_FIELD");`
if(terms.hasPositions()){
TermsEnum termsEnum = terms.iterator();
PostingsEnum postings = null;
while(termsEnum.next() != null){
postings = termsEnum.postings(postings ,PostingsEnum.ALL);
while(postings.nextDoc() != PostingsEnum.NO_MORE_DOCS){
System.out.println(postings.nextPosition());
}
You need to do some of your own analysis to arrive at the data that you need but your first need to save meta data as pointed in point # 1.
}
}
searcher is IndexSearcher instance, hit.doc is doc id and hit is a ScoreDoc .
I have Lucene index which has city names.
Consider I want to search for 'New Delhi'. I have string 'New Del' which I want to pass to Lucene searcher and I am expecting output as 'New Delhi'.
If I generate query like Name:New Del* It will give me all cities with 'New and Del'in it.
Is there any way by which I can create Lucene query wildcard query with spaces in it?
I referred and tried few solutions given # http://www.gossamer-threads.com/lists/lucene/java-user/5487
It sounds like you have indexed your city names with analysis. That will tend to make this more difficult. With analysis, "new" and "delhi" are separate terms, and must be treated as such. Searching over multiple terms with wildcards like this tends to be a bit more difficult.
The easiest solution would be to index your city names without tokenization (lowercasing might not be a bad idea though). Then you would be able to search with the query parser simply by escaping the space:
QueryParser parser = new QueryParser("defaultField", analyzer);
Query query = parser.parse("cityname:new\\ del*");
Or you could use a simple WildcardQuery:
Query query = new WildcardQuery(new Term("cityname", "new del*"));
With the field analyzed by standard analyzer:
You will need to rely on SpanQueries, something like this:
SpanQuery queryPart1 = new SpanTermQuery(new Term("cityname", "new"));
SpanQuery queryPart2 = new SpanMultiTermQueryWrapper(new WildcardQuery(new Term("cityname", "del*")));
Query query = new SpanNearQuery(new SpanQuery[] {query1, query2}, 0, true);
Or, you can use the surround query parser (which provides query syntax intended to provide more robust support of span queries), using a query like W(new, del*):
org.apache.lucene.queryparser.surround.parser.QueryParser surroundparser = new org.apache.lucene.queryparser.surround.parser.QueryParser();
SrndQuery srndquery = surroundparser.parse("W(new, del*)");
query = srndquery.makeLuceneQueryField("cityname", new BasicQueryFactory());
As I learnt from the thread mentioned by you (http://www.gossamer-threads.com/lists/lucene/java-user/5487), you can either do an exact match with space or treat either parts w/ wild card.
So something like this should work - [New* Del*]
I come up with solution to programmaticlly create query to search for phrase with wildcards using this code:
public static Query createPhraseQuery(String[] phraseWords, String field) {
SpanQuery[] queryParts = new SpanQuery[phraseWords.length];
for (int i = 0; i < phraseWords.length; i++) {
WildcardQuery wildQuery = new WildcardQuery(new Term(field, phraseWords[i]));
queryParts[i] = new SpanMultiTermQueryWrapper<WildcardQuery>(wildQuery);
}
return new SpanNearQuery(queryParts, //words
0, //max distance
true //exact order
);
}
Example creation and call toString() method will output:
String[] phraseWords = new String[]{"foo*", "b*r"};
Query phraseQuery = createPhraseQuery(phraseWords, "text");
System.out.println(phraseQuery.toString());
outputs:
spanNear([SpanMultiTermQueryWrapper(text:foo*), SpanMultiTermQueryWrapper(text:b*r)], 0, true)
Which works great, and fast enough for most cases. For instance, if I create such query and search with it, It will output desired results, for example:
Sentence with foo bar.
Foolies beer drinkers.
...
And not something like:
Bar fooes.
Foo has bar.
I have mentioned that query work fast enough in most cases. Currently I have an index with size of aprox. 200GB and on average searching time is between 0.1 to 3 seconds. Depending on many factors like: cache, size of subsets of documents matching single word in phrase since lucene will perform set intersections between founded terms.
Example:
Let supose I want to query phrase "an* karenjin*" (which I will split into ["an*", "karenjin*"] and than create query using createPhraseQuery method) and I want that it matches sentences containing: "ana karenjina", "ani karenjinoj", "ane karenjine", ... (different cases due croatian grammar).
This query is very slow that I haven't waited long enough to get results (over 1h) and sometimes causes GC overhead limit exceeded exception.
This behaviour is somewhat expected since "an*" itself matches a huge number of documents. I am aware of that I could query "an? karanjin*" which giver results in 30-40sec (faster but still slow).
This is where I am confused.
If I query just "karenjin*" it gives results in 1 sec. Therefore I have tried to query "an* karenjin*" and using a Filter "karenjin*" using WildcardQuery and QueryWrapperFilter. And it is still unacceptable slow (I killed process before it returned anythong).
Documentation says that Filter reduces search space of Query. So I tried to use filter:
Filter filter = new QueryWrapperFilter(new WildcardQuery(new Term("text", "karanjin*")));
And query:
Query query = createPhraseQuery(new String[]{"an*", "karenjin*"}, "text");
Than search, (after several warm-up queries):
Sort sort = new Sort(new SortField("insertTime", SortField.Type.STRING, true));
TopDocs docs = searcher.search(query, filter, 100, sort);
OK, what is my question?
How come is quering:
Query query = new WildcardQuery(new Term("text", "karanjin*"));
is fast, but using Filter described above is still slow?
Yes, wildcards can be performance hogs, especially if they match a lot of terms, but what you describe does seem surprisingly so. Hard to say for sure why that is occuring, but for an attempt.
I'll assume:
Query query = new WildcardQuery(new Term("text", "an*"));
On it's own, is performing very badly, as described. Since the wildcards you are looking for are both prefix style queries, it's a better idea to use a PrefixQuery instead.
Query query = new PrefixQuery(new Term("text", "an"));
Though I don't think that will make much of a difference if any at all. What might just make a different is changing you rewrite method. You could try limiting the number of Terms the query is rewritten into:
Query query = new PrefixQuery(new Term("text", "an"));
//or
//Query query = new WildcardQuery(new Term("text", "an*"));
query.setRewriteMethod(new MultiTermQuery.RewriteMethod.TopTermsRewrite(10));
I would like to use Lucene to index/search text. The text can contain mistyped words, names, etc. What is the most simple way of getting Lucene to find a document containing
"this is Licene"
when user searches for
"Lucene"?
This is only for a demo app, so we need the most simple solution.
Lucene's fuzzy queries and based on Levenshtein edit distance.
Use a fuzzy query in the QueryParser, with syntax like:
Lucene~0.5
Or create a FuzzyQuery, passing in the maximum number of edits, something like:
Query query = new FuzzyQuery(new Term("field", "lucene"), 1);
Note: FuzzyQuery, in Lucene 4.x, does not support greater edit distances than 2.
Another option you could try is using the Lucene SpellChecker:
http://lucene.apache.org/core/6_4_0/suggest/org/apache/lucene/search/spell/SpellChecker.html
It is a out of box, and very easy to use:
SpellChecker spellchecker = new SpellChecker(spellIndexDirectory);
// To index a field of a user index:
spellchecker.indexDictionary(new LuceneDictionary(my_lucene_reader, a_field));
// To index a file containing words:
spellchecker.indexDictionary(new PlainTextDictionary(new File("myfile.txt")));
String[] suggestions = spellchecker.suggestSimilar("misspelt", 5);
By default, it is using the LevensteinDistance, but you could provide your own customized Edit Distance.
I have a database table with about 40,000 records containing code fields, such as
FLEFSU25B-25M
EMG1090-5S
I need to be able to very quickly select all codes that contain a given substring. For example "109" matches EMG1090-5S.
My current approach is to store the codes in Lucene and have Lucene filter by substring - such as 109
But that is not very efficient if I just store the codes, because than Lucene has to search through all the tokens.
To overcome this, I'm thinking of creating a new analyzer that will split each code into tokens, like this:
EMG1090-5S
MG1090-5S
G1090-5S
1090-5S
...
Then to find all codes with substring 109, I can search on 109* which is much more efficient (I understand Lucene stores tokens alphabetically, just like SQL Server indexes).
Does this make sense?
Does such an analyzer already exist? I'm using .Net/C#.
A token filter to accomplish this does indeed already exist! Take a look at EdgeNGramTokenFilter. An Analyzer using it might look something like:
Analyzer analyzer = new Analyzer() {
#Override
protected TokenStreamComponents createComponents(String fieldName, Reader reader) {
KeywordTokenizer source = new KeywordTokenizer(reader);
LowercaseFilter filter = new LowercaseFilter(source);
filter = new EdgeNGramTokenFilter(filter, EdgeNGramTokenFilter.Side.BACK, 2, 50);
return new TokenStreamComponents(source, filter);
}
};
For your consideration, WordDelimiterTokenizer might also prove useful to you. It has a number of configuartion options, and can be used to separate at punctuation and at transitions from letter to number, etc. So with it, you could get the from your input: "EMG1090-5S"
You could get the tokens:
EMG
1090
5
S
Which might work well for your case, but would not be particularly helpful in finding something like: "MG1"