org.terrier.matching.daat

Class Full

java.lang.Object

org.terrier.matching.BaseMatching

org.terrier.matching.daat.Full

All Implemented Interfaces:

Matching

Direct Known Subclasses:

FatFull

public class Full
extends BaseMatching

Performs the matching of documents with a query, by first assigning scores to documents for each query term and modifying these scores with the appropriate modifiers. Documents are matched in a document-at-a-time fashion. In particular, the posting lists for all query terms are processed in parallel (but without threads). In comparison to TAAT matching, this reduces the memory consumption during matching, as documents which will not make the final retrieved set are discarded. After matching, the document score modifiers are applied if necessary. This Matching strategy uses the PostingListManager for opening and scoring postings.

Since:

3.5

Author:

Nicola Tonellotto and Craig Macdonald

See Also:

PostingListManager

Field Summary

Fields inherited from class org.terrier.matching.BaseMatching

collectionStatistics, documentModifiers, dsmNamespace, IGNORE_LOW_IDF_TERMS, index, invertedIndex, lexicon, logger, MATCH_EMPTY_QUERY, numberOfRetrievedDocuments, queryTermsToMatchList, resultSet, RETRIEVED_SET_SIZE, totalTime, wm

Constructor Summary

Constructors

Constructor and Description
Full(Index index) Create a new Matching instance based on the specified index

Method Summary

Methods

Modifier and Type	Method and Description
protected void	assignScore(int i, CandidateResult cc) assign the score for this posting to this candidate result.
String	getInfo() Return a human readable description of this Matching class
protected void	initialisePostings(MatchingQueryTerms queryTerms)
protected CandidateResult	makeCandidateResult(int currentDocId)
protected CandidateResultSet	makeResultSet(Queue<CandidateResult> candidateResultList)
ResultSet	match(String queryNumber, MatchingQueryTerms queryTerms) Get a ResultSet for the given query terms.
protected int	selectMinimumDocId(it.unimi.dsi.fastutil.longs.LongPriorityQueue postingHeap) returns the docid of the lowest posting

Methods inherited from class org.terrier.matching.BaseMatching

finalise, getTotalTime, initialise, setCollectionStatistics, updateEndTime, updateStartTime

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Full

public Full(Index index)

Create a new Matching instance based on the specified index

Method Detail

initialisePostings

protected void initialisePostings(MatchingQueryTerms queryTerms)

Overrides:

initialisePostings in class BaseMatching

match

public ResultSet match(String queryNumber,
              MatchingQueryTerms queryTerms)
                throws IOException

Get a ResultSet for the given query terms.

Specified by:

match in interface Matching

Specified by:

match in class BaseMatching

Parameters:

queryNumber - - some ID of the query

queryTerms - - query terms to match

Returns:

ResultSet - the matched results

Throws:

IOException - if a problem occurs during matching

makeResultSet

protected CandidateResultSet makeResultSet(Queue<CandidateResult> candidateResultList)

makeCandidateResult

protected CandidateResult makeCandidateResult(int currentDocId)

assignScore

protected void assignScore(int i,
               CandidateResult cc)
                    throws IOException

assign the score for this posting to this candidate result.

Parameters:

i - which query term index this represents

cc - the candidate result object for this document

Throws:

IOException

selectMinimumDocId

protected final int selectMinimumDocId(it.unimi.dsi.fastutil.longs.LongPriorityQueue postingHeap)

returns the docid of the lowest posting

getInfo

public String getInfo()

Return a human readable description of this Matching class

Specified by:

getInfo in interface Matching

Specified by:

getInfo in class BaseMatching